I build the layer between a working model and production.
nyu distributed systems ml infra llm platforms
flint NL to DAG compiler. plain english in, running pipeline out.
orbis-ml xlm-roberta entity classifier · 120+ countries · 98.75% precision
falcon production ML inference · nginx · multi-worker · exponential backoff
audit-ai clinical NLP for healthcare compliance
puneethkotha.com · linkedin · python kafka fastapi pytorch redis docker k8s
