I work on getting models to actually run in production
mostly working on async pipelines, inference servers, and messy real-world data
ml infra distributed systems llm systems
flint - NL to DAG compiler for ML pipelines
orbis-ml - xlm-roberta classifier with multi-stage pipeline across 120+ countries
falcon - inference server with FastAPI, nginx, multi-worker setup
audit-ai - finds compliance issues in clinical text



