Andrei Pokhilko
Andrei works in the CTO office at Komodor as Open Source Dev Leader with 20+ years of engineering experience spanning performance testing leadership at Blazemeter (Chief Scientist), startup co-founding, and strategic technical roles. He's the creator of multiple successful open source projects including JMeter-Plugins (transformed into ecosystem marketplace), Taurus, Helm Dashboard, and Komoplane. As a passionate advocate for knowledge sharing, Andrei has been speaking at conferences worldwide for 15+ years, covering topics from performance testing to Kubernetes and AI. He believes deeply in building technology that helps others succeed.
Session
This hands-on workshop guides participants through building a complete ML pipeline using production-grade tools on Kubernetes. Starting with a Kind cluster, you'll implement each component of a modern MLOps stack: Airflow for orchestration, Spark for data processing, Ray for model training and serving, and MLflow for experiment tracking. Participants will create a pipeline that processes multiple dataset sizes, trains variant model architectures, and automatically deploys the best performer. Perfect for developers and data scientists looking to bridge the gap between ML experimentation and production deployment.