Remote – Regular – Engineering
- The ML Data team enables rapid development of features and training/evaluation datasets. We provide a batch and real-time feature store, a central ML Dataset Store, as well as a signal platform with monitoring and governance tools. The team works with Spark, Parquet, Flink streaming, as well as Java online services, and Python for orchestration.
- The ML Training team builds large-scale training, model management/deployment tools, and modeling libraries. We build a Training Compute Platform provisioning GPU and cluster compute, model deployment, and PyTorch ML modeling libraries providing cutting-edge model building blocks. The team works with Python, MySQL, and full-stack UI development, as well as interacting with Kubernetes, Spinnaker, Jenkins.
- The ML Serving team builds our universal engine for large-scale ML model inference in online, offline, and streaming contexts. We also develop services to fetch features and provide visibility/monitoring/observability on deployed ML systems. The team works with C++ online serving and hardware optimization.
What you’ll do:
- Design and build core components of the ML lifecycle
- Work with internal customers, ML engineers and data scientists across Pinterest, to understand and solve development velocity pain points
- Work with many major ML teams in Pinterest to productionize their models, improve their productivity and unblock ML innovations
Job information can change without notice
Share with an amigo to support the website