Show HN: Serverless platform for inference of time-series foundation models

Hi, I’m Andrei — We’ve built FAIM, a serverless platform for running inference with foundation time-series models like Chronos2, TiRex, and FlowState. It’s built on Nvidia Triton and GCP, so you can get forecasts in seconds without managing any infrastructure.

We built it because we believe the lack of accessible infrastructure for foundation time-series models is one of the main reasons these models haven’t yet spread widely in industry.

To try it out, register on https://faim.it.com/ to get your API key, then run models serverlessly via our Python SDK: https://github.com/S-FM/faim-python-client.