Inside Lila's AI Research Platform
summary
Lila Sciences runs a self‑service, cloud‑native AI platform that claims to handle multicluster, multicloud, multiregion workloads and to shrink ideation time. It uses pyflight run (Flight v1) to build and containerize, Ray for in‑cluster fan‑out to avoid the 10,000 pods anti‑pattern, a Chariot CLI that centralizes auth, launches and logs, and Hydra as the Hydra source of truth for configs. Policy enforcement uses Kyverno and OPA, and pipelines already process thousands of sequences and fold hundreds of structures successfully.
AI-written summary · about
chapters (6)
- 0:00Introduction to Lila's AI platform
- 3:58Platform goals and requirements
- 9:37Software stack and architecture
- 16:41Onboarding friction and solutions
- 21:17Self-service wins and examples
- 24:50Q&A with the team
related talks
17mDevOps for Private AI: A Scalable Platform Using Kubeflow
42mLLMs on Autopilot: Running AI Agents on Kubernetes With Open Source Tools
27mGen-AI at Scale: Simplifying Orchestration of Healthcare Applications Across Multi-Cluster Environme
21mMachine Learning and Microbiology in the Cloud Through Ray
27mCloud Native Artificial Intelligence - Top to Bottom
38mAI in Kubernetes: How to Get Started?
33mFrom Chaos To Control: Building ML Platform
35mBuilding a Medical AI with Kubernetes and Kubeflow