Full-Stack Engineer specializing in AI integrations, multi-agent systems, and robust backend infrastructure.
I build full-stack AI applications. My work focuses on bridging reliable systems engineering with practical LLM integrations—specifically multi-agent orchestration, local AI environments, and robust RAG pipelines. I prioritize shipping clean, working software over hype.
A multi-agent platform designed to automate customer support across 6+ channels (Slack, Discord, Email, etc.). Features real-time WebSocket monitoring and persistent conversational memory. Successfully reduced manual response workload by 85%.
A diagnostic tool for the Anthropic Model Context Protocol (MCP) ecosystem. The engine establishes Stdio/SSE connections to analyze server health, validate configurations, and automatically generate production-ready setups.
A cross-platform CLI tool that transforms any standard USB drive into a portable, offline AI environment. Automates the deployment of Ollama and AnythingLLM runtimes without leaving a data footprint on the host machine.
- Models: Gemini 2.0/2.5, Claude 3.5 Sonnet, GPT-4o, Local LLMs (Ollama)
- Frameworks: LangChain, LangGraph, CrewAI, AutoGen
- Data & RAG: Vector DBs (Qdrant, ChromaDB, FAISS), Graph Databases
- Infrastructure: Python, Rust, FastAPI, TypeScript, Next.js, Node.js, Docker, GitHub Actions
- Agentic AI Integration: Deep practical experience building autonomous workflows and AI-driven applications.
- MCP & LLM Workflows: Advanced implementation of Anthropic’s Model Context Protocol for automated coding workflows.
- AI/ML Intern @ DeveloperHub: Developed and shipped 11+ production-level prototype applications.
- BS Software Engineering (Est. 2028): Academic foundation in systems architecture and engineering principles.
- Location: Lahore, Pakistan
- Email: mu.ai.dev@gmail.com
- Portfolio: buildwithusman.me
- LinkedIn: muhammad-usman-ai-dev