Python · FastAPI · ML · LLM · CUDA · Rust
Systems-first ML engineer who ships production inference, data pipelines, and fast backend services.

• Production-grade backend services and APIs with Python and FastAPI for model serving and tooling.
• Reliable data ingestion and ETL for analytics and retrieval augmented workflows.
• LLM infra: prompt iteration, eval loops, lightweight retrievers and RAG orchestration.
• Performance work: CUDA-aware inference optimizations and low-latency Rust sidecars.

• Built FastAPI backends that expose model outputs, instruments, and realtime endpoints used by product teams.
• Designed and ran ingestion pipelines and ETL that feed OLAP and retrieval systems.
• Implemented evaluation and ranking loops to measure and improve LLM output quality.
• Wrote Rust watcher/updater services for low-latency node monitoring and state propagation.

• FamAgent — modular agent orchestration with a FastAPI backend.
  https://github.com/kaushal07wick/FamAgent
• Finsight / sqlagent — financial analytics engine and Airflow + FastAPI tooling.
  https://github.com/kaushal07wick/sqlagent
• BreeBoost — real-time fraud detection pipeline with drift monitoring and alerting.
  https://github.com/kaushal07wick/breeboost
• LanceDB contributions — ingestion and vector search integration work.

Python · FastAPI · AsyncIO · SQL · Postgres · ClickHouse · Docker · Airflow · CUDA · PyTorch · vLLM · Rust · Git · CI/CD · Logging · Metrics

• Make model outputs production-ready with clear APIs, validation, and observability.
• Turn prototyped LLM flows into repeatable RAG pipelines.
• Reduce latency and cost through targeted inference optimizations.
• Ship small, high-impact features fast in early-stage teams.

Email: [email protected]
Twitter: https://twitter.com/ofcboogeyman LinkedIn: https://www.linkedin.com/in/11kaushalkumar/