Research Engineer
Atul Saxena
I build the environments models are trained in, and the tools that show what they learned.
- RL Systems
- Interpretability
- ML Infrastructure
Research Engineer
Atul Saxena
I build the environments models are trained in, and the tools that show what they learned.
Calibrated decisions in milliseconds, self-hosted
An open decision model for agents and pipelines. Give it a state and typed questions; get back calibrated probabilities, an abstain signal and an uncertainty measure. It never writes text.
Using a chat model as a decision-maker means generating a sentence and then parsing it — slow, brittle, and with no calibrated confidence. There is also no clean way for it to say "none of these" rather than picking the least-bad option.
status: Released · open weights