Projects

Things I've built

Side projects where I test an idea end to end — and write up what worked and what didn't. All built on public or simulated data.

FeaturedAgents · Trading · 2026

AI Fund House

A multi-agent paper-trading fund. AI agents act as analyst, risk manager and CIO; code enforces the risk limits; I approve every trade.

Three agents with separate roles that challenge each other
Risk limits checked in code, not left to the model
A human approves every trade

View case study Read the build guide

Reinforcement learning · Finance · 2020 → 2026

From trading bot to trading agents

My MSc reinforcement-learning trading bot, revisited: what changes when the agent can reason in language.

Write-up coming soon

ESG · Public data · In progress

Why ESG ratings disagree

Looking at Hong Kong listed companies where rating providers diverge — and which methodology choices explain the gap.

Follow along in my notes →

Trend following · Backtesting · In progress

The great traders, tested with data

Turning the entry and exit rules of Livermore, O'Neill, Minervini and Qullamaggie into code, and backtesting them across asset classes to see which signals could anchor a quant fund.

Follow along in my notes →

Earlier work

Research projects from my MSc in Artificial Intelligence (2020).

Reinforcement-learning trading agentAn agent that learns when to buy, hold or sell from daily price and indicator data, tested against classic moving-average strategies on Sharpe ratio and drawdown.GitHub → Deep-learning price signalsTurning technical indicators (RSI, MACD, moving averages) into trading signals with a sequence model, with backtesting you can rerun from GitHub.GitHub → Faster image segmentationCombining a fast object detector with a segmentation network as a lighter alternative to the standard, slower approach.GitHub →

All code on GitHub →

Have an idea worth testing?

I'm open to collaborating on AI and data projects in finance and ESG.

Send me a message