I build AI-powered tools and data pipelines. I came to software through data — started with machine learning and IBM certifications, then got serious about Python fundamentals and modern AI development workflows. Now I ship across the full stack: from exploratory data analysis to production-deployed AI apps with live URLs.
Currently completing Dave Ebbelaar's Python for AI course while building real projects inspired by his AI Cookbook.
- Self-Improving Agent — Claude critiques and rewrites its own output across iterations until quality is reached — live demo ↗
- Agentic RAG — Claude searches documents using tools (list, grep, read) with live reasoning steps — no pre-embedding — live demo ↗
- Memory Agent — AI assistant that remembers you across every conversation using Mem0 — live demo ↗
- AI Email Client — production PWA that brings Claude AI into your inbox across Gmail, Office 365, and IMAP — live demo ↗
- YouTube Transcript Analyzer — Python MCP server + web app for transcript analysis with Claude AI — live demo ↗
- RAG Document Q&A — full-stack app to chat with your own documents using OpenAI and Next.js — live demo ↗
| Project | Description | Stack | Live |
|---|---|---|---|
| self-improving-agent | Agent that writes a draft, critiques it (score + weaknesses), rewrites, repeats — feedback loop pattern | Next.js · TypeScript · Claude API | ↗ |
| agentic-rag | Claude uses list_files / grep / read_file tools to search docs — live reasoning panel with citations | Next.js · TypeScript · Claude API | ↗ |
| mem0-memory-agent | AI assistant with persistent long-term memory across sessions powered by Mem0 Cloud | Next.js · TypeScript · Claude API · Mem0 | ↗ |
| ai-email-client | AI-first email PWA — Gmail, Office 365 & IMAP with Claude AI summaries, reply drafts and prioritisation | Next.js · TypeScript · Claude API | ↗ |
| python-mcp-automation-claude-desktop | Python MCP server for Claude Desktop + web app for YouTube transcript analysis | Python · FastMCP · Next.js · Claude API | ↗ |
| docling-rag | RAG document Q&A — upload any PDF or URL, ask questions, get cited answers | OpenAI · Next.js · TypeScript | ↗ |
| Project | Description | Stack |
|---|---|---|
| cpg-forecasting-eda | M5 Forecasting EDA — CPG sales analysis and visualisation | Python · Pandas · Jupyter |
| ibm-python-final-project-loan-analysis | IBM ML capstone — loan repayment prediction with SVM, KNN, decision tree and logistic regression | Python · scikit-learn |
| ibm-gdp-data-extraction | IBM capstone — global GDP data extraction via web scraping and Pandas | Python · BeautifulSoup |
| Project | Description |
|---|---|
| python-for-ai | Exercises and projects from Datalumina's Python for AI course |
| ai-engineer-notes | Study notes, cheat sheets and references for AI engineering |
Languages
AI & Agents
Web & Infra
Tooling
IBM Python for Data Science, AI & Development
IBM Machine Learning with Python
IBM Developing AI Applications with Python and Flask
↓
EDA, Pandas, scikit-learn — M5 Forecasting, GDP extraction, loan prediction
↓
Python for AI by Datalumina — fundamentals, uv, Ruff, project structure
↓
6 production AI apps — RAG, Agentic RAG, MCP servers, memory agents,
feedback loops, AI email client (Next.js + Claude API + Vercel)
↓
Next: multi-agent systems · LangGraph · production ML pipelines
Building in public from Perth, WA.