Six builds spanning regulated advisory tooling, multi-agent trading systems, and classroom software — each one built to solve a problem I ran into directly, not a portfolio exercise.
A multi-agent crypto trading parliament — Momentum, MeanReversion, Sentiment, RiskGuardian, OrderFlow and an LLM reasoning agent voting through a consensus engine — settling trades on Kraken with Circle USDC nanopayments on Sepolia testnet, fed by yfinance and Alpaca market data.
A South African personal finance planning platform built to CFP® planning standards — client and advisor portals, retirement projections, tax calculations, and an FSCA-aligned AI assistant, backed by Supabase and built with FAIS and POPIA compliance in mind from the first commit.
An M&A intelligence platform built around six coordinated AI agents that research targets, model valuations, and surface diligence flags — applying the same DCF and comparable-company instincts from my HP Tech Ventures work to a faster, automated first pass.
A spaced-repetition study app built on the SM-2 algorithm, with a Focus Lock mode to block distractions during study sessions, folder-reading for turning existing notes into review material, and Groq/Llama integration for AI-assisted card generation.
A live classroom game platform built for UFS Industrial Psychology students, running on a Supabase realtime backend so lecturers can run synchronous quiz and engagement sessions without leaving the browser. Deployed and used in real classroom sessions.
A CRM built for South African financial advisors — client management, built-in SA tax calculators, and a Claude AI integration for drafting client communication, built to shorten the gap between a client conversation and a compliant record of that conversation.