← Journal · · Announcement
We launched Selectel's Inference platform
Hey, hey, heeey! Yesterday at Tech Day we announced our new product, which I'm directly involved in developing: the Inference platform! The platform is built on nvidia triton server, which you've probably seen in my Habr articles quite a bit. And it really is a great open source inference server that closes a ton of pains around optimizing its work at the hardware level.
I talked about its main features in a talk that's already available via the link. This time it's more product-focused, since the conference is about our products, not the tech. But it still covers why we built this platform and its main functionality :) So go check it out!
But if you're curious about what problems we ran into and how we solved them while building the platform, I'm happy to tell you that in December I'll be talking about the technical details at HighLoad++ There I'll finally tell you why we moved away from Seldon, how we implemented autoscaling, canary deploy, and the inference graph 🔥🔥🔥