Aditya Raut

Meet Aditya

1 following

I make RAG systems work in production — not just demos.
Cut retrieval latency 40%, token costs 60%, ingestion from 70s→27s at my current internship. Merged PR in Haystack (25k+ stars). Full stack: PyTorch, ONNX, TensorRT, FastAPI, Docker, edge deployment on Jetson Nano.
If you need a RAG pipeline built, an LLM app optimized, or a model actually running in your infra — I've done it for real.