I'm a researcher working on AI systems for science and engineering, with interests in RL infrastructure, test-time scaling, CUDA kernels, self-improving AI systems, and long-horizon scientific and software engineering tasks. My research focuses on AI for semiconductor manufacturing, autoresearch, and AI for multiphysics and inverse design. On this site you can find my work and my open-source projects on GitHub.
What I tinker with day to day.
Building
Ultra-Fast AI Superfactory for Science. Agent-native science needs a continuous factory on your GPUs. Give Mythein Agent the outcome you want; it runs the experiments, simulations, and training that turn a question into evidence—so your research moves faster.
Founding Engineer
AI for HVAC systems and energy products.
Research notes, build logs, and the occasional deep dive.
A Codex plugin that hands one task to Claude Fable, Opus, Sonnet, Haiku, or Sakana Fugu without leaving the session, on the subscriptions already on your machine.
A verifier-driven continual learning recipe that combines self-generated traces, replay, and a failure-driven curriculum to learn new task families while limiting forgetting.
A compact, agent-driven research harness for Mamba-2, Mamba-3, and Nemotron-H-style hybrid Mamba–Transformer MoE models on MLX and CUDA.
An open-source catalog of 19 agent-maintained skills spanning autonomous research, LLM post-training, GPU/TPU programming, and accelerated computing.
A memory layer for Recursive Language Models that records trajectories, extracts reusable strategies, and lifts a weaker model's benchmark score by 26% over three rounds.
How a coordinated agent team researched, synthesized, and integrated a production-grade React Native skill into a shared knowledge base in under 15 minutes.