@ISHESATOSHI · BENGALURU, IN
@isHeSatoshi
I'm an LLM research engineer. I work on agents, evaluation harnesses, and inference: improving pretrained autonomous-driving models with reinforcement learning, shipping production RAG, and distilling models down until they run on a consumer GPU.
Most of my time goes into measuring things rather than guessing: how attention and quantization actually behave at runtime, whether an eval set is really holding out what it claims, and where the architecture breaks before production finds it.
Currently working on
- LLM agents
- Agent evaluation
- Inference & quantization
- RAG pipelines
- Continual learning
- Model distillation
- Computer vision
- Edge & embedded