Developing ML models is not a novelty anymore. The focus shifted towards optimisation and ability to perform advanced operations ideally from one place. This demo will present an end-to-end machine learning pipeline. It consists of essential steps required for development of ML models. Pipeline will present the steps from the beginning of the process to the deployment. It will include loading the dataset, pre-processing the data, model training, model versioning and model deployment. We will demonstrate the whole process with open source ML workflow tool Charmed Kubeflow running in Kubernetes cluster. During this demo, the audience will learn about the open source MLOps platforms and how it enables optimized training for ML modeling.
Session author bios
Michal Hucko is an MLOps engineer at Canonical working for the Kubeflow team. He has a rich experience in building ML platforms in cloud, end to end ML pipelines and in general machine learning model deployment process. Michal loves to talk about programming and machine learning, thanks to his passion in teaching others how to code.
Bartłomiej Poniecki-Klotz is the AI/ML Engineer in Canonical where he works with clients on tailor-made solutions for their AI/ML, Data Platform and HPC projects. In the past, Bartlomiej designed and implemented MLOps platforms for F500 companies. He built highly scalable computation platforms for banks and led AI/ML team.Thanks to the Software Engineer and Data Science background, Bartlomiej builds tools for Data Scientists to focus on experimentation and MLOps Engineers to transform experiments into sustainable AI-powered applications.
|Level of Difficulty||Beginner|