Interview
summary
Built a production machine learning platform at Netflix that gives data scientists on-demand large-scale compute early in development and removes concerns about scaling models and data pipelines, so teams focus on experiments not ops. Platform design follows use-case driven platform decisions: add features only after seeing workflow needs. Work includes exploring AutoML hyperparameter optimization and using Scala for developer ergonomics; the platform is in active use by data science teams.
AI-written summary · about
related talks
29mEvolving the ML Platform Organisation at Netflix: A Case Study
03mInterview
04mInterview
53mKeynote Interview
09mPlatform Engineering for Data Science at Scale
18mBuilding a Machine Learning Platform at
32mRebuilding the AI Platform: Designing for Today’s Chaos and Tomorrow’s Unknowns
42mThe Challenges Managing a Kubernetes-Based Machine Learning Infrastructure