Syed Muhammad M.

I build AI systems that run where your data lives: LLM agents and RAG on your own servers, and computer vision on NVIDIA Jetson at the edge. Top Rated, 100% Job Success, 6 years building production AI. Recent work: • Multi-agent HR assistant: LangGraph supervisor routing to SQL, reporting and HR agents through an MCP tool server, with prompt-injection and SQL guardrails, role-based login, Docker and Kubernetes. • DevOps copilot: an LLM agent that debugs pods and opens GitHub pull requests on real Kubernetes clusters over MCP, with human approval on every change. Runs on OpenAI or local models (Ollama, vLLM). • Natural-language SQL for Oracle/PeopleSoft: plain-English questions turned into safe, read-only SQL across a 6,000-table database. • Edge vision in production: vehicle analytics on Jetson Xavier, ATM security monitoring, AI dashcams, multi-camera monitoring on Jetson Orin, drone swarm ground control. What I can do for you: • LLM agents and automation: LangGraph, LangChain, MCP servers, tool calling, guardrails • RAG and document Q&A: hybrid search, re-ranking, Qdrant / FAISS / Pinecone • Private, offline AI: local LLMs for data that can't leave your network • Edge AI on NVIDIA Jetson: Nano, Orin and Xavier setup, TensorRT optimisation, RTSP video pipelines • Computer vision: YOLO, OpenCV, PyTorch for detection, tracking, counting and face recognition • Backends and deployment: FastAPI, Django, PostgreSQL, Redis, Docker, Kubernetes, AWS Lambda I'm currently co-founder of Mecha Innovations, where we build on-premise AI assistants for companies. Expect clear scope, regular updates and clean, documented code. Message me with what you're building and I'll reply with a plan.