"A passionate Software Engineer who hates the 'Vibe-Coding' term — instead tries to use AI to build software using proven computer science techniques."
A multi-agent orchestration framework built around a written constitution — the H-Factor Protocol — that separates planning from execution from review across agents, with a Peer Review gate that blocks a step from running until it's approved. On top of that: visual DAG pipelines, an 11-agent registry (Claude Code, Codex, Gemini, Hermes, and more), cascading execution with topological (Kahn's-algorithm) ordering, and a single Next.js dashboard covering 63 API endpoints, 18 pages, token/cost tracking, and an evidence-based 110-question code benchmark.
A documented collection of Claude Skills — reusable, self-contained workflows that extend what Claude can do automatically:
| Skill | What it does |
|---|---|
| 📝 Notes OCR | Handwritten scans → Markdown + Excel + PDF, with faithful Mermaid diagram reconstruction |
| 💡 Brainstorm-to-Code | Interactive Q&A → clean Python class skeletons |
| 🧠 Agent_Brain | A bio-inspired agent architecture (cerebral-cortex lobes → orchestration modules), with 10 UML diagrams |
A personal knowledge base / Obsidian vault cataloguing the AI development stack — languages, frontend, databases, infra, AI providers, and agentic harnesses — organized as a structured reference for the full dev toolchain.
Built by Abdullah Abdul Hakim · Dubai 📍 · @7akim97 · LinkedIn