← Journal · · Announcement
Announcement: talk at the Ostrovok DevOps meetup!
Hi everyone! This Thursday, March 27 at 18:30, I'll be in Moscow giving a talk on Inference Autoscaling in K8s at the Ostrovok DevOps meetup! This time the meetup covers everything from serious senior-level talks to devops stand-up comedy, so it should be interesting for a broad audience.
In my talk I'll show how we autoscale Triton on our inference platform. I already talked about this at HighLoad++, but this time I decided to expand the topic into a standalone talk - added scale to zero, scale from zero, and how we fight autoscaling bottlenecks through image caching and faster pulling.
The talk will be interesting not just for ML folks but for devops specialists too, since inference autoscaling isn't that different from autoscaling web services overall.
Come in person or join online! The meetup is free, registration is available via this link