Trent Richards
Strategy major building AI tools and autonomous systems across business, technology, and product.
Projects
AI day trading bot scanning 50+ tickers every 30 minutes. A multi-agent debate system — bull, bear, and risk agents argue each position using Llama 3.3 via Groq — then votes resolve into BUY/SELL/HOLD with bracket orders and GTC stop losses. Runs live on Alpaca paper trading.
Interactive skill graph with 75+ nodes across two layers — a public map of technical skills and a private brain tracking beliefs, goals, and interests. D3.js force-directed visualization with distance-based color saturation. Python pipeline ingests Claude conversation history and Obsidian vault notes to auto-update node scores and connections, then commits to GitHub via API.
Minimal time tracker built around a single design constraint: it should feel good to use. One-tap clock in, full-screen focus mode styled after the iPhone lock screen, 7-day stacked bar charts. No accounts, no backend — everything in localStorage. A deliberate exercise in building useful tools with clean, opinionated UI.
Strategy major who builds. Most people in one world don't speak the other — I'm working on being the exception.
I got into software because I think the next few years will require people who can actually make things with AI, not just manage the people who do. So I started building: a trading bot with a multi-agent debate system, an interactive skill graph that auto-updates from my work history, a time tracker designed around how focus actually feels.
Just finished a semester as an AI research consultant at PwC.
The goal is systems that generate value on their own — and eventually, the strategic sense to know which ones are worth building.
Languages
- Python
- TypeScript
- Cantonese
AI
- Ollama
- Groq / Llama
- RAG
- Embeddings
Stack
- Next.js
- React
- D3.js
- Tailwind
- Docker
- Git
- Vercel
- SQLite