Choose timezone
Your profile timezone:
Update: The rooms are the following:
Be a part of it!
Are you astonished by what one can accomplish with free and open-source software? Are you amazed by what open-source communities put together with great enthusiasm? Would you like to know how this all relates to our daily life?
Would you also like to become one of these individuals? Are you interested in coding for one of the many great free software projects? Applications? Desktop environments? System components? Cloud? IoT? OS Distribution integration? Immutable OS distribution?
But also as non-programmer you are highly welcome for documentation, design, web, community, ...
And all this does not only give you the good feeling of helping in the community and making free software better, it also raises your chances on the job market, mentioning your contributions and accomplishments in your resume, especially for the numerous positions in the area of open source ...
So come and join us! In this conference you will learn a lot, about projects, coding, documentation, ... There will be talks, panels, Q&A, interactive workshops, BoFs/meet-ups, demos, ... and lots of great hallway sessions ...
We will cover many great topics (more to come):
The actual subject matters also depend on the submissions we get on our Call for Abstracts. We are eager to see your amazing ideas! Note that you cannot only submit talks and workshops here, but also ideas for demos, hackathons, and evening events. Your submissions will make up the conference.
This is our third Opportunity Open Source. Here is how the first two were:
We are on Mastodon: #OpportunityOpenSource Follow this hashtag to stay imformed!
Red Hen Lab has recently been doing work at the intersection of natural and artificial intelligence, specifically exploring how natural intelligence not only comes up with uniquely appropriate solutions, but does so with a far higher degree of efficiency than the current iterations of AI. Early in the history of computing, a handful of large corporations and government institutions imagined the future of computing as a time share service. The current vision of AI is similar: a set of highly advanced tools dominated by a handful of megacorporations controlling and monetizing these tools as a service.
We suggest that what is needed is an international microbiome of open-source projects that leverage our emerging understanding of the efficiency of natural intelligence to develop a new generation of tools. These tools will undercut the lumbering giants of AI with their multi-billion dollar budgets and dedicated nuclear power stations. The challenge is to generate machine learning strategies and tools that require a fraction of the resources used by current models. All the way from birth, humans are capable of one-shot observational learning, keeping training costs at a minimum. By developing categorical and hierarchical models of the world, we can rapidly prospect possible solutions and their consequences without having to search indiscriminately through astronomically large combinatorial spaces. Implementing such strategies in AI has the potential to dramatically lower training costs and democratize the development of AI tools.
Open Source is more than just code—it’s a global movement built on collaboration, transparency, and shared growth. For newcomers, however, the first step can feel overwhelming: Which project should I choose? Do I need to be an expert? How do I make my contributions count?
This talk will demystify the process of getting started with Open Source. We’ll cover the essentials—from understanding community culture and finding beginner-friendly projects to making your first contribution and building a sustainable journey. Whether you’re a student, professional, or hobbyist, you’ll learn practical strategies for identifying opportunities, collaborating effectively, and showcasing your work.
By the end of this session, you’ll walk away with a clear roadmap, actionable resources, and the confidence to take your first (or next) step in the world of Open Source.
Getting selected for an LFX Mentorship feels like winning a lottery - but it's not about luck. It's about strategy, persistence, and understanding what makes a strong application stand out among thousands of applicants.
In this talk, I'll take you through my complete LFX mentorship journey - from the moment I first heard about the program to receiving that life-changing acceptance email from KubeEdge. This isn't just another success story; it's a practical roadmap based on real experience.
I'll share the exact steps I took to research and choose the right project, how I crafted my application to demonstrate genuine interest and capability, and the mistakes I made along the way that you can avoid. You'll learn about the application timeline, what mentors actually look for in candidates, and how to make your proposal stand out in a competitive pool.
We'll dive into what the LFX mentorship experience is really like - the weekly check-ins, the learning curve, the challenges of contributing to a complex project like KubeEdge, and how it's transforming my understanding of open source development. I'll also discuss the technical skills you need, the soft skills that matter just as much, and how to prepare yourself to be a valuable mentee.
KubeEdge, being at the forefront of edge computing and Kubernetes, presents unique challenges and opportunities. I'll share insights into the project's ecosystem, the community dynamics, and what it's like to work on cutting-edge technology that's shaping the future of distributed computing.
This talk is perfect for students and early-career developers who are curious about LFX but don't know where to start, those who have applied before and been rejected, or anyone wanting to understand how to break into major open source projects through structured mentorship programs.
By the end, you'll have a clear action plan for your own LFX application, realistic expectations about the program, and the confidence to take that first step toward becoming an open source contributor.
Open source is powering the next wave of AI innovation, but the real challenge lies in making AI infrastructure accessible, secure, and scalable across environments—from cloud-native platforms to the edge devices.
In this talk, we’ll begin with a high-level perspective on the role of open source in shaping AI adoption across industries. From there, we’ll dive into practical architectures that enable AI to run efficiently in the cloud leveraging Kubernetes, exploring how virtual clusters (vCluster) provide secure multi-tenancy for teams while ensuring resource efficiency. We’ll connect this with running AI on-device, highlighting how AI can extend beyond datacenters to devices at the edge, enabling real-world applications that demand low latency and privacy-preserving computation.
We will focus on the next frontier of AI in making it personal, private and accessible. Attendees will walk away with a clearer understanding of the opportunities and challenges in running AI at scale using open source—whether in the cloud, or at the edge.
EunoiaMind is an open-source, emotionally intelligent AI framework designed to support mental wellness, built entirely on a free-tier infrastructure with no GPU, no cloud dependencies, and one working hand using ubuntu and HuggingFace free-tier.
This session presents EunoiaMind not just as a chatbot, but as a foundational framework that combines affective computing, cognitive-behavioral therapy structures (CBT, ACT, DBT), accessible design, and narrative echo modeling - all built using ubuntu and open-source tools like Hugging Face, Gradio, and SentenceTransformers.
We will explore how this system was designed from scratch with empathy and constraint at its core: developed by a disabled technologist, awarded nationally for work in accessible tech and rooted in over a decade of work in speech recognition and open-source contribution.
Immutable Linux distributions are talked about a lot. Ease of use and security by the operating system itself and also the applications in read-only file systems and isolated from each other, this an idea coming from smartphone operating systems.
Well before the immutable hype, more than a decade ago, Canonical, having given up on the smartphone operating system Ubuntu Touch, based on their experience with Touch, they created the immutable system Ubuntu Core, and Snap as packaging system for apps and components.
Later on, they established Snap as universal, distribution-independent packaging system for classic Linux distributions and the Snap Store started offering all kinds of applications, including desktop applications.
Out of that arose Canonical's approach of an immutable desktop distro, Ubuntu Core, with a desktop environment Snap added, Ubuntu Core Desktop.
This talk is about how all this works, its motivations, advantages, challenges, and state-of-the-art ...
Till Kamppeter and Aveek Basu are organizing every year the participation of the Linux Foundation as mentoring organization in the Google Summer of Code (GSoC) and are mentoring contributors for OpenPrinting.
In all the years (17) we did not only get a lot of work done, but got people into our team, got contributors into the world of free software, gave them experience for their careers, ...
In this panel session Till and Aveek, and some of the current and former contributors and mentors will tell about their experience with GSoC, what was amazing, where they struggled, ...
We tell how to get a GSoC contributor, what is required and how one works on a project. We also show the evaluation and onboarding process of OpenPrinting ...
And we do a Q&A session for the audience, answering all their questions and dispel any doubts ...
And if you have experience with the GSoC, as a contributor/student and/or as a mentor/org admin, please feel free to step up and join us in our panel on the stage!
Many government and business organizations in Russia are switching from Windows to GNU/Linux right now, or have already done so. This big countrywide move has caused strong growth in Linux development inside Russia. It also means Russian Linux distributions, like ROSA Linux, are being adapted to meet complex needs of large enterprises.
ROSA Linux distribution is a sovereign distribution, not based on any other distribution, the continuation of the Mandriva distro. It is one of TOP-4 Linux distributions in Russia for both home and business use. It also has one of the largest collections of software packages in the world (see repology.org). Since 2012, ROSA Linux has been built and maintained using ABF (abf.rosa.ru) — a self-developed Automated Build Farm — which manages source code of packages in git, builds them, publishes them into testing and release repositories and builds installation images (ISOs).
ROSA supports multiple hardware architectures:
ROSA is also the base for other distributions, providing commercial support to partners building custom distros on its package base. For example, MOS — a ROSA-based distro, that has its own instance of ABF (abf.mos.ru) and contributes to its mother repository — ROSA. ROSA provides an enterprise-ready and stable Linux platform (repository + ABF + easy-to-use ISO build scripts), while MOS builds an OS based on it. MOS is used in all schools of Moscow (> 150 000 installations).
This massive switch to Russian-made Linux in big companies and government offices – probably one of the largest migrations of its kind globally – has forced adaptations for enterprise use. GNU/Linux has already been a good and easy-to-use desktop OS, but home users do not need enterprise-specific features like, for example:
This talk will cover these and other problems and our experience in solving them. Distribution-specific features will also be convered, for example:
I will also tell how we balance between stability and fresh software.
Modern JavaScript runtimes aren’t just tools—they’re mini operating systems. Like kernels, they manage resources, isolate programs, and define how code runs.
In this talk, we’ll explore how Deno uses syscall-inspired design using Rust to build a sandboxed execution environment. We'll also touch on asynchronous I/O tasks like file access, subprocesses, and networking works and how the event loop schedules tasks under the hood.
Conference attendees to Till:
Printing works better than under Windows or Mac!
Michael Tunnell, TuxDigital, says in one of his videos:
There is no such thing like a pain-free experience of printing under Windows ... Linux printing is ridiculously good ...
Till Kamppeter, leader of the OpenPrinting project, and fellow of the Linux Foundation, will give an overview of his work.
Going through OpenPrinting's history the components of the printing infrastructure of modern Linux (and other Posix-style) operating systems will get shown.
Especially also the New Architecture of all-IPP printing and scanning and also the integration in immutable OS distributions will be treated ...
In the end also Microsoft's move to go a similar all-IPP way without classic printer drivers, Windows Protected Print, is also shown.
OpenPrinting web site
Monthly News
Destination Linux Interview
This work presents a practical and efficient approach to accelerating the development of the Scilab Control System Toolbox through AI-assisted workflows. The focus is on reimplementing key control system functions from GNU Octave within the Scilab environment, thereby enhancing native capabilities of Scilab.
Traditionally, the development of a single function, including code implementation, test case creation, and documentation, could require up to two weeks. With the integration of AI-assisted tools, this process can be reduced to a matter of hours, enabling significantly faster development while maintaining reliability and functional accuracy.
The methodology involves analyzing existing Octave functions, generating corresponding Scilab functions using AI, followed by expert review, validation against Octave outputs, and refinement where necessary. While AI expedites the development process, human oversight remains essential to guide the workflow and ensure correctness, as AI-generated code may contain errors or incomplete logic.
To date, approximately 50 out of a targeted 150 functions have been successfully reimplemented using this approach. The work contributes to the open-source community by broadening access to advanced control system tools in Scilab and demonstrates the potential of combining AI capabilities with domain expertise to accelerate scientific software development. The same methodology can be extended to other toolboxes and domains within the open-source ecosystem.
Keywords: Scilab, GNU Octave, Control Systems, AI-assisted Development, Open Source Software, Scientific Computing
Now, after Till has given you an overview about OpenPrinting, you probably have many questions, for example:
Till, Aveek, Alexander, Akarshan, Mohammed (and perhaps other people from OpenPrinting) will answer your questions.
AWS's S3 (Simple Storage Service) is one of the most popular public cloud storage offering and for a reason. But public cloud offerings are designed to lock organisations in a suite of propitiatory solutions which results in a stacked up bill fast. In this workshop we would understand the principles of cloud object storage (S3) and then develop a demo app that uses S3 served by MicroCeph Object Gateway (RGW). The attendees should carry a laptop, and should be exposed to python programming.
Project Description
The ScaniVerse project is a pioneering initiative designed to streamline the installation and management of scanner drivers across different Linux distributions. By utilizing Snaps, the project enables the creation of distribution-independent packages, allowing scanner drivers to be easily installed on any system that supports snapd. Additionally, scanner applications can be containerized using OCI containers, facilitating their deployment on immutable operating systems. Key components of the project include the integration of scanner support into the PAPPL library, originally developed for printer applications. This integration supports multi-function devices and provides a unified driver format for both printers and scanners. The project also focuses on retrofitting legacy scanners to ensure continued support for older hardware.
Open source software (OSS) is popular used and underpins critical infrastructure, which is often defaultly installed or runs transparently without end users consciously recognizing dependencies. Ensuring the security and stability of these OSS components is therefore essential for systemic trust and resilience. Existing methods for evaluating the security of OSS, such as OpenSSF Scorecard, which provide automated, project-based assessments, but such approaches remain coarse-grained and fail to capture nuanced aspects of a project’s developer context.
In this talk, we introduce a novel OSS Security Evaluation Metrics framework that integrates developer‑centered behaviors, community dynamics by combining human‑centric observations, and the help of thriving large language models. The framework yields richer, contextualized insights into OSS security posture. We validate this approach via a comprehensive benchmark across a diverse set of OSS ecosystems, revealing key gaps in current tooling and providing a stronger foundation for trustworthy open‑source ecosystems.
Point-of-care diagnostics represents a critical frontier in healthcare delivery, particularly in resource-limited settings where traditional diagnostic infrastructure is often inadequate or inaccessible. Despite significant advances in artificial intelligence and edge computing technologies, existing medical diagnostic solutions predominantly rely on single-modality analysis and cloud-based processing, creating substantial gaps in accessibility, privacy protection, and real-time decision support for healthcare providers operating without reliable internet connectivity or expensive infrastructure. This research addresses these critical limitations by developing a comprehensive multi-modal AI framework that integrates open-source large language models with computer vision capabilities, specifically optimized for deployment on low-cost Raspberry Pi platforms to enable simultaneous analysis of textual clinical data and medical imaging without internet dependency. The proposed system incorporates federated learning mechanisms to ensure privacy-preserving collaborative model improvement while maintaining local data control, advanced model compression techniques to achieve real-time inference on resource-constrained hardware, and offline training capabilities that eliminate dependency on cloud services. The outcomes include a significant reduction in diagnostic costs, improvement in diagnostic accuracy through multi-modal data fusion, a decrease in time-to-diagnosis, and the creation of an open-source framework that democratizes access to sophisticated medical AI technologies while building local technical capacity in underserved healthcare environments worldwide.
Three students, one shared dream, and countless late-night coding sessions - this is the story of how we transformed from complete open source newcomers to LFX Mentees at KubeEdge, one of the most exciting edge computing projects in the Kubernetes ecosystem.
Just months ago, we were typical computer science students with no idea how to contribute to open source projects. The world of GitHub issues, pull requests, and community discussions seemed intimidating and exclusive. But through collaboration, persistence, and a commitment to learning together, we not only overcame these barriers but all three of us secured prestigious LFX mentorships with KubeEdge.
In this talk, I'll share our authentic journey - the mistakes we made, the breakthroughs that changed everything, and the practical strategies that worked. You'll learn about our systematic approach to understanding complex projects like KubeEdge, how we built our technical skills while contributing meaningfully to the community, and most importantly, how supporting each other made all the difference.
We'll explore what KubeEdge is and why edge computing represents the future of distributed systems, but through the lens of newcomers discovering these concepts for the first time. I'll also break down the LFX mentorship application process, sharing the insider tips we learned and the common pitfalls to avoid.
Whether you're a student wondering how to make your first contribution, someone curious about edge computing and Kubernetes, or simply looking for motivation to start your open source journey, this talk will show you that with the right approach and community support, amazing opportunities are within reach.
By the end of this session, you'll have a clear roadmap for getting started in open source, understand what makes projects like KubeEdge special, and hopefully feel inspired to begin your own journey with friends by your side.
FOSS technology solutions are undoubtedly the cornerstone of the most important technological trends globally. When it comes to artificial intelligence, cybersecurity, software development, blockchain, and digital transformation in general, FOSS technologies are optimal.
The creation of business models around projects developed using FOSS technologies offers an enormous opportunity to consolidate technological development in public, private, and even citizen-based settings by leveraging agile project management methodologies such as Scrum and Canva.
The integration of FOSS technologies and agile methodologies presents an excellent opportunity to generate highly robust and viable technological projects, derived from the management of multidisciplinary collaborative teams. Leveraging FOSS technologies offers the advantage of establishing a service-based business model free from the enslaving nature of expensive licensing and membership schemes.
Fuzzing is a powerful testing technique for uncovering hidden bugs, crashes, and security vulnerabilities. But for many developers, getting started with fuzzing and integrating it into real-world projects can feel overwhelming. In this workshop, you will learn how to use OSS-Fuzz, Google’s infrastructure for continuous fuzz testing, to secure open source software.
Drawing from real experience integrating OpenPrinting projects into OSS-Fuzz during Google Summer of Code 2025, this workshop will guide you through every essential step:
By the end of the session, you will have a working fuzzing setup, a running fuzzer, and a better understanding of how to secure your software through automated testing.
Who should attend?
This workshop is ideal for students, open source contributors, and developers who want to learn practical fuzzing. No prior experience with fuzzing is required, but basic knowledge of programming and using Docker will be helpful.
This research presents a novel approach to obstacle detection during navigation using a
combination of Convolutional Neural Networks (CNNs) and Long Short-Term Memory
(LSTM) networks. The primary objective is to generate accurate image captions that describe
the content of images, which is crucial for applications such as autonomous driving and
assistive technologies for the visually impaired. We systematically analyze the architecture of
our model, which consists of three main components: a CNN for feature extraction, an LSTM
for sequence generation, and a mechanism for sentence formulation. By employing transfer
learning with the Inception v3 architecture, we enhance the model's performance while
reducing computational costs. Our experiments utilize the Flickr8k dataset, which comprises
8,000 images, each accompanied by five descriptive sentences. We introduce a simplified
version of Gated Recurrent Units (GRUs) as an alternative to LSTMs, demonstrating
comparable performance with fewer parameters, thus improving training efficiency. The
model's effectiveness is evaluated using the Bilingual Evaluation Understudy (BLEU) score,
which quantifies the quality of generated captions against reference sentences. Results indicate
that our architecture achieves a BLEU score of aprox 80% on the training set and approx 75%
on the test set, showcasing its capability to produce semantically and grammatically correct
captions. Additionally, we explore the integration of attention mechanisms to enhance the
model's focus on relevant image features during caption generation. The findings suggest that
our approach not only meets the challenges of automatic image captioning but also holds
potential for broader applications in image understanding and navigation systems. Future work
will involve expanding the dataset and refining the model to further improve accuracy and
robustness in diverse scenario
The evolution and consolidation of open source technologies offer an extraordinary alternative for driving technological innovation and digital transformation. Today, open source is the cornerstone of major global technological trends. The freedom to fully utilize it and the catalyzing of generational talent create an extraordinary environment for the development of fantastic projects.
Snap and Docker have become foundational tools in the open-source ecosystem, offering simplified distribution, cross-platform compatibility, and reproducible environments. Their strengths are particularly evident in scenarios where traditional software deployment is hindered by OS fragmentation, dependency conflicts, or limited user privileges—conditions commonly encountered in academic and resource-constrained environments.
Despite these advances, many domain-specific open-source tools—especially in technical fields like Electronic Design Automation (EDA)—remain difficult to deploy and scale across such diverse systems. One such tool is eSim(https://esim.fossee.in/home), an open-source EDA suite developed by the FOSSEE project(https://fossee.in) at IIT Bombay. It integrates components like KiCad, Ngspice, OpenModelica, and Verilator to support schematic design, simulation, and PCB layout—offering an open alternative to proprietary software like OrCAD or LTspice.
However, eSim’s adoption has often been limited by its platform-specific dependencies and complex installation workflows. To bridge this gap, we packaged eSim using Snap and Docker, enabling frictionless, single-command installation on multiple Linux distributions and reproducible, offline-compatible execution environments. Snap offers automatic updates and dependency bundling across distros, while Docker enables eSim to run in controlled lab setups or low-connectivity environments without requiring root access.
Through this talk, we share our experience containerizing eSim, discuss technical challenges we resolved, and reflect on how such approaches can dramatically increase the usability and impact of domain-specific open-source tools. Our work highlights how container technologies are not only infrastructure solutions but also key enablers of educational access and sustainable software adoption.
DSA Visualizer is an open-source, interactive web platform that brings Data Structures and Algorithms (DSA) to life through real-time visualizations, code execution tracking, and memory simulation. Built at IIIT Kottayam by the student-led Openverse club, it aims to make algorithm learning intuitive, engaging, and fully accessible to all learners.
This is more than a visualizer. It is a comprehensive DSA learning environment, featuring:
Hash Tables, Recursion, Graphs, Search, and Greedy Algorithms
Interactive quizzes across all major topics to test conceptual understanding
Topic-wise curated LeetCode problems, neatly sorted for guided progression
Live search functionality across visualizers and problem banks
| Category | Tools & Libraries |
|---|---|
| Frontend | React 19, Vite 6 |
| Animations | Framer Motion |
| Code Display | Monaco Editor, React Syntax Highlighter |
| Design System | Custom dark-mode UI |
| Backend (Future) | flask, local storage |
| Hosting | IIITK self Hosted (soon) |
Code logic is primarily implemented in C, in sync with IIITK’s DSA121 course.
This project is built with the belief that CS education should be visual, interactive, and open to all. DSA Visualizer is being developed as a free educational resource — made at IIITK, by IIITK, for everyone. We invite educators, developers, and students worldwide to contribute, adapt, and expand the platform.
Live Demo: https://visualalgo-ov.web.app
Github Repo: https://Openverse-iiitk/DSA-Website
Made with 💚 by Openverse, IIIT Kottayam
Virtualization at the Core: Trends, Hypervisors, and the Shift to Cloud-Native Platforms
Abstract Virtualization continues to be the bedrock of modern IT infrastructure, enabling flexible, efficient, and scalable environments. This paper outlines key trends in virtualization, explains the role of hypervisors, and provides a concise understanding of how platforms like OpenStack and OpenShift build upon these technologies. It aims to present a 360-degree view of how virtualization continues to evolve and support emerging computing paradigms.
I'll talk about Office Gob MX and its implementation in the Mexican government using Ubuntu as the operating system. I'll also explain how the document validator we developed works.
Free and Open-Source Software (FOSS) has had a profound impact on the software industry, enabling unprecedented collaboration and innovation. It empowers developers, fosters a spirit of sharing, and accelerates the development process by leveraging existing open-source components. Businesses strategically adopt FOSS to gain brand exposure and leverage the collective effort of the open-source community. The joy of sharing and the collective contributions of developers and maintainers form the backbone of successful FOSS projects.
But what happens when this utopian vision meets reality?
While FOSS brings numerous benefits, it also casts long shadows. This talk will venture into the "dark side" of FOSS, exploring the critical challenges that threaten the very ecosystem we celebrate.
Have you already thought about how the applications you develop get distributed to end users?
Often developers only provide the source code. So for not so tech-savvy users the major distributions need to pick up projects and package them, and maintain them for new releases.
This is why there is a need for distribution-independent, secure, and easy-to-use packaging, like on smartphones. This exists also for Linux ... Among the options, there is Snap! Applications are easy to find in the Snap Store and they are installable on most Linux distributions.
More about Snap: The Powers, The People
https://snapcraft.io/
In this workshop we introduce you to the Snap packaging format and guide you on how to create your first Snap. Hands-on exercises are done with 3 simple applications. To be well prepared to do the exercises please download the slides (Under "Presentation Materials" below) already before coming to the conference/the workshop and follow the instructions in the "Setup" section.
So if you are creator/developer/maintainer of an application and want to reach a wider audience with it, make it easier to install for end users, you will learn what you need to do to get your work ready for the Snap Store here.
Even if your application is not a desktop application, command line applications, system applications and daemons, ... could be packaged as Snaps, too.
Also if you want to help snapping applications whose upstream developers do not have snapped them, you should also attend this workshop and join the Snapcrafters.
We present an offline AI assistant that uses open-source tools to help with the design and debugging of electronic circuits. The system works with eSim, an Electronic Design Automation (EDA) software that was made as part of the FOSSEE project at IIT Bombay by integrating various Open Source tools like KiCad, Ngspice, GHDL, Verilator and many more. This program encourages students and teachers in resource-limited settings to use free and open-source software to get around problems they face. It is aimed at people in areas with poor internet access and schools like tier-3 colleges that can't afford expensive proprietary EDA tools. This makes it possible for everyone to have access to high-quality circuit design workflows.
Traditional EDA workflows, on the other hand, often include complicated toolchains, hard-to-understand simulation errors, and a lot of manual work to fix them. These problems are even worse for new designers and students who don't have access to commercial support. Our assistant solves these problems by giving automated, easy-to-understand help throughout the whole design process.
The assistant uses a modular pipeline that works well on low-end hardware and doesn't need an internet connection to work. An automated watcher service keeps an eye on file changes in popular tools like KiCad and Ngspice. When it sees a change, it parses netlists and error logs to keep design states and diagnostics up to date. OpenCV is used for visual support, which lets users monitor schematics based on images, recognize components, and check layouts to find and fix design mistakes early on. The user interface is meant to work well in both terminal-based and graphical environments. It is designed to work with the most common Linux distributions used in schools.
The assistant uses a retrieval-augmented generation (RAG) framework to give high-quality, context-aware answers when people talk to it in natural language. A vector-search retrieval system keeps track of local documentation and domain-specific knowledge bases. This lets the query generation module put together structured prompts. The Qwen 1.8B language model processes these, which are hosted locally by the Ollama framework. The fact that it works completely offline protects users' privacy and lets them use chain-of-thought reasoning to get detailed, step-by-step explanations for fixing designs, figuring out simulation errors, analyzing logic, and evaluating circuits. A session-based memory module keeps conversations going, so users can build complex designs step by step with the same help each time.
The assistant makes advanced circuit design workflows easier to access, faster, and easier to debug by combining multimodal analysis, semantic retrieval, and fully offline language-model inference into one workflow. This helps make modern circuit design and prototyping more accessible to everyone by lowering design mistakes, speeding up development cycles, and helping students and teachers learn new skills when they can't afford expensive tools or have reliable internet.
Keywords: Electronic Design Automation, offline AI, circuit debugging, eSim, Qwen LLM, OpenCV, KiCad, Ngspice, RAG, Circuit Analysis, RTL code analysis, Linux UI, session-based memory, chain-of-thought reasoning.
In today’s digital ecosystem, enterprises demand more than just a content management system — they expect intelligent, adaptive, and personalized experiences. Drupal has long been recognized as a flexible and powerful CMS, but with the rise of Artificial Intelligence (AI), it is evolving into a central hub for orchestrating AI-driven digital experiences.
This session will explore how Drupal can act as the “intelligent layer” in enterprise architectures by integrating AI capabilities such as generative content creation, personalization, translation, accessibility, and workflow automation. We will discuss real-world use cases where Drupal leverages AI/ML services from open-source frameworks and cloud providers (like Azure AI, Google Vertex, and AWS AI) to create experiences that are not only content-rich but also context-aware.
By rethinking Drupal as an AI-powered Digital Experience Platform (DXP), organizations can unlock new opportunities to scale, innovate, and differentiate in a crowded digital landscape. This talk is designed for architects, developers, and digital leaders looking to future-proof their Drupal implementations in the era of AI.
Key Takeaways
Understand how AI transforms Drupal from a CMS into a DXP hub.
Learn practical AI integrations with Drupal: content generation, personalization, moderation, and translation.
Explore enterprise-ready architectures combining Drupal with AI services across multi-cloud environments.
Gain insights into governance, compliance, and ethical use of AI in open-source ecosystems.
Discover future directions where Drupal and AI together can redefine digital experiences.
Target Audience
- Drupal Architects & Developers
- Enterprise Digital Leaders
- Open Source AI Enthusiasts
- Product Owners / CTOs exploring AI adoption
Many grand challenges like climate change and pandemics emerge from complex interactions of millions of individual decisions. While LLMs and AI agents excel at individual behavior, they can't model these intricate societal dynamics.
Enter Large Population Models LPMs: a new AI paradigm simulating millions of interacting agents simultaneously, capturing collective behaviors at societal scale. It's like scaling up AI agents exponentially to understand the ripple effects of countless decisions.
AgentTorch, our open-source platform, makes building and running these massive simulations accessible. It's optimized for GPUs, allowing efficient simulation of entire cities or countries. Think PyTorch, but for large-scale agent-based simulations. LPMs are already making real-world impact. They're being used to help immunize millions of people by optimizing vaccine distribution strategies, and to track billions of dollars in global supply chains, improving efficiency and reducing waste.
In this talk, we'll dive into the underlying technology of LPMs, provide an overview of AgentTorch, and explore how you can contribute to both our research at MIT and the open-source project. We'll also discuss intriguing topics like prompting LPMs versus LLMs, opening up new avenues for AI development. Whether you're interested in complex systems, AI, or tackling global challenges, LPMs offer exciting opportunities for innovation and impact.
AI-native applications are poised to redefine how users interact with technology — but for AI to be truly seamless, scalable, and privacy-respecting, it must move closer to the user. On-device AI is the key to unlocking adaptive, real-time experiences while eliminating the latency, cost, and privacy trade-offs of cloud-based models
In this talk, we’ll dive into how developers can harness DeliteAI, an open-source on-device AI platform, to create intelligent applications that dynamically adapt to each user.
You’ll learn how to:
We'll walk through real-world patterns, code examples, and performance benchmarks to show how on-device AI can enable magical, responsive experiences that weren't previously possible.
By the end of this session, you'll have a practical framework for integrating on-device AI into your mobile apps — unlocking AI for everyone, everywhere.
Ready to take your coding game to the next level?
Join us for a hands-on workshop where you’ll experience Vibe Coding — coding with flow, creativity, and real-world best practices.
You’ll get to work with Saarthi, our AI-powered VS Code Dev Team that supports you through the entire SDLC — from writing clean code to testing, reviewing, and deployment. Plus, you’ll learn how to apply Godspeed Framework’s best practices to build scalable and maintainable software like industry pros.
What you’ll gain:
Hands-on practice with agentic coding using Saarthi
Software development best practices made simple
Exposure to real-world framework used by top teams
A chance to vibe-code with peers and build confidently.
Free and open-source software is meant to be for everyone. But that promise often falls short when digital barriers are built into our interfaces. Accessibility (a11y) is not an afterthought—it's a foundational design principle that determines who gets to participate in the open-source ecosystem.
This session will demonstrate how to integrate accessibility into the frontend development process from day one. We’ll walk through practical strategies for building inclusive components using React, Tailwind CSS, and Storybook. Attendees will learn how to test their interfaces with tools like axe-core, eslint-plugin-jsx-a11y, and Lighthouse, and how to design UI patterns that serve users with diverse abilities.
We'll also explore the human side of a11y in open source: writing inclusive documentation, reviewing pull requests with empathy, and encouraging contributors from all backgrounds.
By making accessibility part of your everyday workflow—not an extra task—you help build software that reflects the values of the community: openness, equity, and collaboration.
As printers and their drivers increasingly become network-aware and cloud-integrated, the need for robust security in printing software has never been more critical. Bugs in such infrastructure components, especially memory corruption or input handling vulnerabilities, can be exploited in real-world attack vectors. Fuzzing, a dynamic testing technique, has proven effective in uncovering such vulnerabilities, as demonstrated by notorious bugs like Heartbleed.
This talk shares hands-on insights from Google Summer of Code 2025, where I contributed to integrating fuzz testing into the OpenPrinting ecosystem. The project aimed to improve the security posture of key open source components: goipp, ipp-usb, pycups, and pyppd, by leveraging Google's OSS-Fuzz platform. These components implement IPP (Internet Printing Protocol) and manage interactions between Linux systems and USB or network printers.
In this session, I will walk through:
This talk is designed for both open source maintainers and students interested in software testing. You'll leave with practical strategies for adopting fuzzing tools, improving code quality, and securing real-world systems through continuous testing. Whether you're just getting started or looking to harden your project against subtle bugs, this session will help you build safer and more resilient software.
This talk will introduce participants to Bitcoin and highlight its significance in today’s world. They will also get to know about Bitcoin FOSS opportunities.
We will first talk about how our current monetary system is broken and the way Bitcoin fixes it. We will also debunk some myths surrounding Bitcoin like the energy consumption issue. Finally we talk about grant programmes supporting Bitcoin FOSS devs and Summer Of Bitcoin.
In this talk, I will share my journey of contributing to open source through Google Summer of Code (GSoC) 2025 under the Linux Foundation’s OpenPrinting project. I will begin by reflecting on how I got started with open source, connected with the OpenPrinting community, and prepared a successful proposal.
The second half of the talk will focus on my GSoC project, “Image Output Evaluation for Testing of Print/Scan Job Processing”, which aims to improve automated quality assurance in print and scan workflows. I will explain the challenges of evaluating printed output beyond crashes and errors, and demonstrate how tools like OpenCV and OCR can be used to verify page completeness, orientation, scaling, color detection, and density analysis.
The session will provide insights into my technical approach, key learnings, and the broader impact this work will have on ensuring reliable and secure printing in open-source systems.
In 2020, Deutsche Bahn's startup accelerator DB mindbox approached BeST Sensor GmbH with the task of developing a monitoring system to detect irregularities in waste water pumps for high-speed trains.
We developed a radically new sensor and signal processing system using RISC edge nodes based on Zephyr RTOS. Several challenges had to be overcome, such as a non-digital legacy infrastructure, harsh operating conditions, extreme quality requirements and a very limited budget. Despite these challenges, we were able to develop a wireless sensor system that permanently monitors the pumping process, automatically evaluates the quality of each process and wirelessly transmits the results to the customer. This system has now been in 24/7 operation for over three years. As a result of continuous improvement, we are now excited to announce the second generation of these devices.
In this talk, I will explain
- How agile project management can work with a large organisation that is often perceived as bureaucratic
- How the open source software community helped us meet seemingly impossible requirements
- how we managed to keep the customer happy and the project within budget
Peephole rewriting forms the backbone of the LLVM optimizer, where over 10% of the size of the codebase of LLVM lives in InstCombine. It's of vital importance to ensure that these rewrites are correct, especially because the canonicalizer and peephole optimizer are run pervasively through the pass pipeline. Efforts such as Alive have brought the verification of peephole optimizations into the LLVM review cycle by providing a convenient user interface for verifying such transformations. However, Alive only checks these rewrites upto a fixed bitwidth (typically, 64), and therefore does not provably guarantee that these rewrites are valid for arbitrary bit-widths.
We take on the challenge of building algorithms that can prove bitvector rewrites correct for arbitrary bitwidths. In particular, these algorithms prove the correctness of any proposition over a fragment of the bitvector theory, with a proof for all bitwidths.
The algorithm is based on ideas from model checking and program deobfuscation. We take care to ensure scalability, and we implement, benchmark, and prove our algorithms correct in the Lean proof assistant. We also provide a user-facing tactic interface for ease of use. . We evaluate our tactic on rewrites from both Hacker’s Delight and the InstCombine test suite, as well as on equivalence checking problems for program obfuscation.
We hope to use our algorithm to simplify the many thousands of lines of proof in the Lean bitvector standard library, and more broadly, to provide effective proof automation for bitvector predicates that are true for all bitwidths.
Dive into the world of real-time operating systems with Zephyr RTOS, the lightweight yet powerful solution for IoT and embedded devices. This hands-on workshop introduces students to the fundamentals of Zephyr, guiding them from installation to building their first application. Perfect for beginners looking to expand their embedded systems knowledge, participants will leave with practical skills in configuring, building, and deploying Zephyr-based applications.
The workshop covers installation of the Zephyr SDK, understanding the build system and working with threads and scheduling. By the end, participants will have practical experience with this powerful open-source RTOS and a solid foundation for building sophisticated embedded systems for the IoT world.
The world is buzzing about the capabilities of large-scale AI models. While proprietary, API-gated systems dominate the headlines, a parallel revolution is empowering developers and enthusiasts worldwide: the open-source AI movement. This is where the true potential for innovation, learning, and building lies for the next generation of engineers. This talk is designed to be your entry point into this exciting world.
In this fast-paced, 20-minute session, we will cut through the noise and provide a practical roadmap for students to start building powerful applications with state-of-the-art open-source models today. We will explore:
The Landscape: A quick tour of the most impactful open-source model families like Meta's Llama, Mistral AI's models, and text-to-image generators like Stable Diffusion. We'll discuss what makes them different and where to find them.
Running Your First Model: We'll demystify the process of running a powerful Large Language Model locally on a consumer-grade laptop, using tools like Hugging Face Transformers and simplified formats like GGUF.
Tuning and Adaptation: Discover how you can make these models your own. We’ll cover the concepts of fine-tuning (training a model on your own data for a specific task) and Retrieval-Augmented Generation (RAG) to build applications like a chatbot that can answer questions about your university coursework.
From Idea to Application: We'll showcase a few compelling, achievable project ideas, such as building a personal code assistant, an intelligent document summarizer, or a unique art generator, to inspire you to start your own AI development journey.
This talk is for anyone who wants to move beyond simply using AI tools and start building with them. You will leave with a clear understanding of the open-source ecosystem, the practical first steps to take, and the inspiration to create your own innovative AI applications.
Prerequisites:
This session is designed for an audience with programming experience (preferably Python) and a basic conceptual understanding of machine learning.
This talk will introduce participants to Bitcoin Design and techniques to make it more accessible. They will also get to know about opportunities in this space.
We will first talk about the intersection of Bitcoin and Design. Then we will move on to accessibility — why it's important and ways to achieve it. Finally we talk about Design opportunities in Bitcoin ecosystem and Summer of Bitcoin.
Venturial is an open source Python toolkit designed to expedite case generation in OpenFOAM, a C++ software for continuum mechanics. The toolkit aims to ease the learning curve of OpenFOAM for new users by offering a flexible workflow with multiple entry points for the user.
The talk will demonstrate Venturial's main features through a sample workflow for generating a mesh for an OpenFOAM case. It will also highlight techniques for modifying the case and conclude with further development aspects to address the needs of intermediate/advanced users.
Over the last few decades, improved hardware and algorithms have expanded general computing capabilities, fueling scientific tools, particularly in computational fluid dynamics (CFD), and making them accessible to a broader audience. One such tool, OpenFOAM, an open-source C++ library prominently used in CFD, is well known for its reusable code design. However, it poses a steep learning curve without a procedural workflow for defining cases. Due to this, new learners often feel discouraged to use this open source software and move to other proprietary software.
OpenFOAM, by itself, is a CLI-based tool that needs a Linux-based Operating System to be used. New learners and people who deal with CFD often get discouraged by these prerequisites for using OpenFOAM. In response to this, Venturial[1], an endeavour by IIT Bombay’s CFD-FOSSEE team[2], envisions to benefit the scientific community by offering a complete CFD workflow with OpenFOAM within python. It provides a user-friendly Graphical User Interface to ease the process of case building for new learners and also provides a Python API to create and modify OpenFOAM cases using a Python script. PyVNT(Python Venturial Node Trees) is a Python API that can be used to read and write OpenFOAM case files and create an abstraction of OpenFOAM case data within Python in a Node Tree-based data structure. Venturial also includes domain-specific presets, proactive rule validation, FOAM-like meshing shortcuts, and personalisation utilities, ensuring a flexible yet structured CFD workflow.
Venturial is inspired by Reynolds-blender[4], a reference implementation of Reynolds[5]. It commenced development in 2021 through FOSSEE’s internship program. Since then, Venturial has been published in SciPy 2024[6], SciPy India 2021 [7], the 18th OpenFOAM workshop [8], and the 2023 IEEE T4E Conference [9].
Our presentation will illustrate Venturial’s meshing workflow using a sample CFD case, emphasising its ease of use for beginners starting with OpenFOAM. We will also showcase PyVNT’s compatibility with OpenFOAM for intermediate and advanced CFD applications. The presentation will conclude with a discussion of Venturial’s future direction, highlighting its student-led development and forthcoming features.
This talk is intended for users and instructors within the CFD scientific community seeking a unified tool for learning or teaching OpenFOAM or developing a customised user interface to interact with various OpenFOAM features. It will provide valuable insights for those aiming to streamline CFD workflows while leveraging a structured and adaptable computational framework.
[1] Adak, Rajdeep, Diptangshu Dey, n.d. FOSSEE/Venturial - A GUI for OpenFOAM. Github: https://github.com/FOSSEE/venturial
[2] “Home,” CFD, https://cfd.fossee.in/
[3] Dey, Diptangshu, Rajdeep Adak, n.d. FOSSEE/PyVNT: Python Venturial Node Trees. Github: https://github.com/FOSSEE/pyvnt
[4] Surti, Deepak, Prabhu Ramachandran, and Shivasubramanian Gopalakrishanan. n.d.
“dmsurti/reynolds-blender: GitHub. https://github.com/dmsurti/reynolds-blender.
[5] Dmsurti (no date) Dmsurti/Reynolds: The preprocessing and solver python toolbox for openfoam., GitHub. Available at:
https://github.com/dmsurti/reynolds
[6] Adak, R., Murallidharan, J. S., & Ramachandran, P. (2024). Venturial: Generating CFD Workflows in Python. Python in Science Conference, 2024 (SciPy), Tacoma, Washington. SciPy. https://doi.org/10.25080/tpwg2365
[7] R. Adak and K. K. Thakur, “Venturial – A Python-based OpenFOAM GUI for CFD Simulations on Blender,” International conference on Python for education and scientific computing, https://scipy.in/2021. pdf available at: https://static.fossee.in/fossee/internship-reports/Python-Blender/Rajdeep_Adak.pdf
[8] Adak, R., Srree Murallidharan, J. and Ramachandran, P. (2023) 18th OpenFOAM Workshop. Available at: https://oxford-abstracts.s3.amazonaws.com/83ca7ab4-c356-4411-be07-070eaeffd43a.pdf
[9] Ramachandran, P., Murallidharan, J. S., & Adak, R. (2023, November 25). Developing Software to Enhance Learnability and Usability of OpenFOAM. Proceedings of International Conference on Technology 4 Education (T4E) 2023 (IEEE T4E), Indian Institute of Technology (IIT) - Bombay. https://doi.org/10.5281/zenodo.14162151
In this talk, I’ll share the story of how we built Plane — a modern, open-source project and product management tool — from the ground up. From our decision to go open source from Day 1 to the way we engage and grow with our community, this session will dive into the principles, processes, and people behind Plane’s development.
I’ll talk about what it really means to build in public, the trade-offs between product direction and community expectations, and the engineering challenges we face daily as developers in a fast-evolving open-source product. You’ll also get a peek into how we plan to sustain our commitment to the open-source ecosystem and ensure that Plane remains a community-first tool.
Whether you're an open-source maintainer, frontend engineer, or someone curious about building developer-focused tools, this talk aims to inspire and inform — with honest lessons from the trenches.
We can all agree that it is important to test the kernel. There are a number of tools for code coverage, dynamic and static analysis to choose from to test the kernel. Linux kernel provides two testing frameworks to write and run tests. A majority of kernel tests are written using these two frameworks. Having a rich set of tools and frameworks is great. However it can be challenging to know which one to chose.
In this talk, Shuah will give an overview of tools and frameworks and their differences, and how to use them to achieve the your kernel testing objectives and goals.
A dynamic demonstration showcasing cutting-edge solutions for the seamless management and deployment of Artificial Intelligence (AI) models within automotive-grade environments. This presentation will highlight an innovative approach to integrating and orchestrating AI capabilities directly on in-vehicle systems.
What You Will Experience:
This session will provide a live, in-depth look at a robust system designed for building, deploying, and managing AI models on AutoSD images. Attendees will gain insights into:
Automated Image Construction: Witness how specialized image building tools can create AutoSD images capable of hosting sophisticated AI models within a QM environment. This streamlines the process of preparing automotive platforms for AI integration.
Secure AI Model Execution: Explore the integration of a powerful runtime environment that enables AI models to operate securely and efficiently within the QM environment, crucial for safety-critical automotive applications.
Centralized AI Model Management: Discover how a distributed system management framework is leveraged to provide comprehensive control over AI models. This includes capabilities for:
Real-time Status Monitoring: Observe the health and status of AI model instances across connected nodes.
Dynamic Model Listing: See how deployed and available AI models can be quickly cataloged and identified.
On-Demand Model Provisioning: Understand the process of pulling and preparing new AI models for deployment.
Interactive AI Inference: Participate in a live demonstration of interacting with a deployed AI model, sending prompts, and receiving real-time responses, showcasing the responsiveness and practical application of the integrated AI.
Key Technical Highlights (Demonstrated Live):
The demonstration will involve a practical walkthrough, illustrating how to:
Prepare and Run the Automotive Image: Building AutoSD images with integrated AI runtime capabilities. Launching and connecting to the live virtualized automotive environment.
Monitor and Manage AI Services: Using command-line tools to inspect the status of nodes and AI-related units. Listing available and deployed AI models within the system.
Deploy and Utilize AI Models: Initiating the download and preparation of new AI models. Activating AI model serving units. Interacting with a running AI model by sending prompts and observing the generated output.
Why This Matters: This demonstration provides a compelling vision for the future of AI in automotive systems, enabling: Faster iteration and deployment of AI-powered features. Enhanced security and isolation for AI workloads. Efficient management of diverse AI models across vehicle fleets. The foundation for true software-defined vehicles with dynamic AI capabilities.
As AI continues to reshape industries, the demand for flexible, secure, and power-efficient computing platforms is rapidly increasing. We DeepComputing built on an innovative RISC-V chiplet-based SoC delivering up to 50 TOPS local AI compute will explore:
- How open RISC-V architecture enables hardware-software co-optimization
- Real-world AI applications in industrial scenarios (chatbots, AGI media, IDEs)
- Lessons learned from bringing an AI PC from prototype to production
Building an AI PC on RISC-V architecture is both a challenge and an opportunity. This year, while developing our latest AI PC, we faced key questions: What AI applications do users truly need? How can we ensure smooth performance on RISC-V laptops? And how can developers effectively leverage local AI computing?
To address these challenges, DeepComputing collaborated with leading open-source communities from Ubuntu. We partnered with one of the most popular open source media players VLC with over 100 million downloads across Linux distributions, optimizing models to run efficiently on local AI computing. Even more wee also try to integrate local AI APIs with Development IDEs like VSCode, and web browsers, using technologies like TensorFlow.js to bring AI to developers right in their workflow.
Growing new talent and attracting new developers is challenging for open source communities. Yet, it is vital to reach out to train the next generation of developers to keep the open source communities healthy and sustainable.
Equitable access to learning resources is a barrier for a significant number of new developers. It isn't easy for new developers to get a start in open source, connect with open source communities and contribute to them. It is equally challenging for employers to find new developers to add to their technical projects.
Shuah Khan will talk about Linux Foundation's six year journey to provide learning resources for new open source developers, opportunities to experts in open source communities to train and mentor the next generation, and make newly trained talent available to prospective employers.
Although driverless printing and scanning are governed by standards and specifications, real hardware implementations often include unique details that can deviate from these specifications, affecting the accuracy of our printing and scanning implementations.
Our years of experience in developing and supporting the Linux driverless printing and scanning stack (CUPS, sane-airscan, ipp-usb) have shown that many issues arise from discrepancies between real-world hardware implementations and the specifications they claim to follow.
Compounding the problem is the fact that printing and scanning hardware is expensive, bulky, and heavy, making it difficult to maintain a representative collection of physical devices for testing.
The solution here would be to implement a realistic hardware simulator capable of emulating not only standard protocol behavior but also device-specific deviations.
This project is currently under active development by OpenPrinting.
The simulator consists of a core engine and a collection of hardware models. A model can be complex, defining every aspect of the hardware’s behavior, or simple, reproducing only the specific issue being investigated.
Each hardware model includes a configuration (represented by parameters such as paper size, supported resolutions, color models, etc.) and optional Python scriptlets that define device-specific behavioral quirks.
The simulation core interprets these models. To simplify model creation, the core provides an "ideal," fully standards-compliant printer or scanner as a base, configurable via the model’s parameters. Small Python hooks then allow deviations from this ideal behavior, which the simulation core automatically applies.
This approach keeps models relatively straightforward. We believe that in cases of printing or scanning issues, system administrators and advanced users will be able to replicate problems using these hardware models, enabling remote troubleshooting without physical access to the device.
The simulation core is written in Go, while Python is used exclusively for hardware models due to its widespread familiarity.
The author specializes in systems programming (C and Go) and is known in the open-source community as the creator of ipp-isb and sane-airscan
Qualcomm Linux is a unified distribution designed to streamline development across a wide range of System-on-Chips (SoCs), integrating key components such as the UEFI bootloader, the latest Long-Term Support (LTS) kernel, and curated drivers supporting popular distributions. This talk will focus on what Qualcomm Linux is and provide an introduction to its software stack and boot flow. We will also explore how Qualcomm empowers customers to build and run their AI applications on upstream distributions with minimal friction, leveraging core hardware functionality. Join us to learn how you can contribute to and benefit from this evolving ecosystem.