SkyPilot AI Infra Meetup at SF TechWeek: GPU Scheduling and SGLang Inference
rseroter · x · 2026-10-09
SkyPilot and H Company hosted an AI infrastructure meetup during SF TechWeek, oversubscribed at 285 spots.
Three technical talks: Seung Jin Yang (SkyPilot) on "Why is my GPU not getting scheduled?"; Léonard Benedetti (H Company) on how they use SkyPilot; and Yuwei An (SGLang core contributor) on SGLang as the unified inference engine for dynamic workflows.
Aimed at infra engineers and researchers building large-scale AI platforms, topics covered pre-training on large GPU fleets, RL infra optimization across training/rollout/reward, and scaling sandboxes to thousands of concurrent agent rollouts.
More from Companies & People
- Musk Announces Grok Bot Users Can Add Shopify Connector via @Bot — elonmusk · 2026-10-09
- Guix bans all AI contributions; dhh fires back: merge good code, reject bad — esrtweet · 2026-10-09
- Waymo co-CEO Dolgov on 20 years of self-driving and why the last 1% is hardest — adityaag · 2026-10-09
- 2026 Nobel Laureate Karl Deisseroth on why he shares science with everyone — ponguru · 2026-10-09
- Mistral slammed for training open models on datacenters powered ~70% by coal — wavefnx · 2026-10-09
- swyx to speak at Forge: why the future is domain-specific everything — sophiamyang · 2026-10-09