Open Source
I build practical Python tools for LLM development, machine-learning environments, and backend engineering.
toolproxy
A provider-agnostic wrapper that adds function/tool calling to LLMs, including models without native tool support. Supports native and emulated tool-call paths, synchronous and asynchronous execution, retries, repair, caching, loop detection, and usage budgets.
GitHub · PyPI · pip install toolproxy
mlenvdoctor
A diagnostic CLI for PyTorch, TensorFlow, and JAX environments. Checks CUDA and driver visibility, dependency mismatches, and generates reports and proposed fixes through a plan–dry run–apply–verify workflow.
GitHub · PyPI · pip install mlenvdoctor
boostapi
A FastAPI scaffolding toolkit with asynchronous SQLAlchemy, Alembic migrations, Redis caching, JWT authentication, rate limiting, and OpenAPI support.