AI Engineer. I build LLM and computer vision systems that survive production.
Currently: Machine Learning Engineer at IF MedTech, working on agentic RAG, eval infrastructure, and multi-model CV pipelines.
What I work on
- Agentic systems: LangGraph, LangChain, MCP, tool-calling agents
- RAG and retrieval: Pinecone, hybrid search, reranking, chunking strategy
- Eval infrastructure: RAGAS, LangSmith, golden datasets, CI regression runs
- Vision: PyTorch, fine-tuning, INT8 quantization, on-device inference
- Full-stack around it: FastAPI, Node.js, Next.js, AWS, Docker, CI/CD, Flutter
Some things I've shipped
- Agentic RAG assistant over 220+ clinical documents; eval harness took recall@5 from 0.70 to 0.87 and faithfulness from 0.72 to 0.89
- On-device fall detection running on 7,000+ devices within 8 months: 83% sensitivity at 1.7 false alarms/day in continuous wear, inference on the MCU with belt and chest mounting and two caregiver-selectable sensitivity profiles
- Multi-model CV inference pipeline serving 270,000+ unique users
- ViT-Base/16 fine-tuned and quantized to ~165 MB for ~0.7s/image on-device
- A CRM platform built solo end to end, live in production
Reach me: X · LinkedIn · mayankbaranwal95@gmail.com