AI & Data Science undergraduate at Bournemouth University, building AI-powered developer tools and agentic systems. I ship small, focused products end-to-end — and increasingly build them with AI agents, not just for them.
🏅 3x AWS Certified (Solutions Architect – Associate) · ✍️ I write about AI on Medium
| Project | What it is | Try it |
|---|---|---|
| Compiler | Turns vague requests into structured prompts, execution plans, and policy-aware workflows — with PR merge-safety checks and MCP tool exports. Offline-first, provider-agnostic. | prcompiler.com |
| VibeGraph | Turn any Python codebase into an interactive call graph with AI-powered explanations — upload, explore, learn. | vibegraph.dev |
| CogniGraph | Natural-language scenario → LLM classification → Brian2 spiking neural network → real-time 3D brain visualization. Educational, not medical. | Live demo |
| NeurIQ | AI-powered personalized learning platform for students — adaptive quizzes, study maps, cognitive analytics. In private beta. | Landing |
| Project | The experiment |
|---|---|
| agy-understudy | A live cockpit for coding agents you delegate to Google Antigravity — watch cheap models do grunt work, in plain language. npx agy-understudy |
| ctx | Zero-dependency "where was I" CLI — snapshot a working directory, resume later with a diff. Built end-to-end by delegated agents, watched live through Understudy. |
- entangled-ai — neural networks learning from each other without sharing data (Medium article)
- stdp_neural — biologically inspired STDP plasticity simulator
- PaperHunterAgent — multi-agent CLI that discovers and summarizes quantum-science papers
- SheetCI — local-first CI guardrail that scans Excel workbooks for risky formulas
- autism-screening-classification — scikit-learn + MLflow screening classifier (educational)
Small, scoped, reversible changes · tests before merge · AI agents for leverage, humans for judgment.