Muhammad Farooq

I'm building provider-agnostic AI coding agents at icod.ai, with pluggable LLM support across Claude, GPT-4o, Ollama, LM Studio, and other OpenAI-compatible models. I developed the AgentLoop, 16 integrated tools, semantic code search, LSP-based code intelligence, and real-time SSE event streaming.

Previously at Phebsoft, I led development of Complimate AI, an agentic compliance platform that generates privacy policies, cookie policies, and Terms & Conditions from business inputs and live website analysis. I built multi-agent pipelines using MCPs and A2A patterns, fine-tuned Mistral and LLaMA on legal corpora, and automated policy generation across 10+ jurisdictions.

My work spans LLM evaluation, context engineering, RAG, vector retrieval, agent safety, and human-in-the-loop controls. I've benchmarked leading LLM providers for quality, latency, and cost while building reliable systems with AWS, Azure, Docker, and CI/CD.

I started in production Python development at Ninesol, where I built FastAPI and Flask APIs, data preprocessing pipelines, and ML-powered analytics services. I also build applied ML projects, including student-performance prediction and VGG16-based chicken disease classification.