Ep168: Scaling Agentic Workloads: Why Reliable Infrastructure is Non-Negotiable for Enterprise AI by Anyscale

** AWS re:Invent 2025 Dec 1-5, Las Vegas - Register Here! **Learn how Anyscale's Ray platform enables companies like Instacart to supercharge their model training while Amazon saves heavily by shifting to Ray's multimodal capabilities.Topics Include:Ray originated at UC Berkeley when PhD students spent more time building clusters than ML modelsAnyscale now launches 1 million clusters monthly with contributions from OpenAI, Uber, Google, CoinbaseInstacart achieved 10-100x increase in model training data using Ray's scaling capabilitiesML evolved from single-node Pandas/NumPy to distributed Spark, now Ray for multimodal dataRay Core transforms simple Python functions into distributed tasks across massive compute clustersHigher-level Ray libraries simplify data processing, model training, hyperparameter tuning, and model servingAnyscale platform adds production features: auto-restart, logging, observability, and zone-aware schedulingUnlike Spark's CPU-only approach, Ray handles both CPUs and GPUs for multimodal workloadsRay enables LLM post-training and fine-tuning using reinforcement learning on enterprise dataMulti-agent systems can scale automatically with Ray Serve handling thousands of requests per secondAnyscale leverages AWS infrastructure while keeping customer data within their own VPCsRay supports EC2, EKS, and HyperPod with features like fractional GPU usage and auto-scalingParticipants:Sharath Cholleti – Member of Technical Staff, AnyscaleSee how Amazon Web Services gives you the freedom to migrate, innovate, and scale your software company at https://aws.amazon.com/isv/

There are no comments yet.

Please log in to write the first comment.