Leveraging Ray to Enhance Machine Learning Models in Finance
summary
Goldman Sachs builds RayLab (kaniko + Kubernetes + Jupyter Lab) to run Ray on private cloud and uses Tune config-driven experiments so research and production share the same model code, enabling reproducibility and easy rollback. Duplicate heavy preprocessing is avoided with a distributed cache keyed by args, letting different nodes reuse results. Large searches are batched (example: five batches) with postprocessing that re-aggregates full-run metrics. Result: faster cluster use, fewer redundant jobs, and deployable real-time yield-curve prototypes.
AI-written summary · about
chapters (6)
- 0:00Introduction to Ray in finance
- 3:38Architectural challenges with Ray
- 8:11Reproducible experiments with Ray Tune
- 12:51Model selection and search strategies
- 16:48Real-time yield curve example
- 20:16Q&A
related talks
31mNext-Generation Data Science Workflows Using Ray
32mRay in Production: Engineering Lessons from Spotify's Hendrix Framework
28mScaling AI & Machine Learning Workloads With Ray on AWS, Kubernetes, & BERT
28mRay on Public-Cloud k8s - experiments, lessons & best practices
44mPanel: ML Platform on Ray
22mRay, a Unified Distributed Framework for the Modern AI Stack
20mThe Quick Journey to Using Ray: How We Implement Ray and Anyscale to Speed up our ML Processes
35mOperating a Production-Scale Physical AI Data Platform with Ray