I am an undergraduate in Electrical Engineering (minor in AI & CS) at the National University of Singapore, supported by the Science & Technology Scholarship. In summer 2026 I studied machine learning and convex optimization at Stanford University through the International Honor Program.
I co-founded Alpha Flow, an early-stage startup building world-model training infrastructure for AI-driven quantitative trading. As CEO, I set the company's direction and guide the architecture of our alpha-discovery systems.
My work sits where machine learning meets markets. Topics I am working on:
- market regime modeling — knowing what state the market is in before deciding what to do,
- world models of market dynamics,
- machine learning for alpha discovery, with strict out-of-sample discipline,
- autonomous agents that run and verify their own experiments.
Projects
Alpha Timing — Regime-Aware Quantitative Trading System
A market-regime classifier routes each trading day to one of four regime-specialist return predictors; a cost-aware dual-threshold rule with volatility targeting sets the position. Out-of-sample on SPY, net of 5 bps costs: Sharpe 1.05 vs 0.69 for buy-and-hold, max drawdown −14% vs −25%.
Regime Classifier — Market-Regime Nowcaster
A supervised RandomForest that classifies the current market state — Bull, Bear, Sideways or Crisis — from trailing price and macro features across 39 US tickers, 2000–2026. Out-of-sample (2019–2026): accuracy 0.877, macro F1 0.846. It is the first stage of Alpha Timing.
ML Doctor — Autonomous ML Research Agent
An agent that runs the full model-improvement loop unattended: it diagnoses from its own scorecard, proposes and justifies a remedy, writes the code, retrains, then checks whether its hypothesis held. I owned the agent brain — a four-role LLM relay and the outer control loop.
Financial AI Agent — Savings Planning System
A full-stack TypeScript monorepo (Next.js, Fastify, PostgreSQL via Prisma) that turns transaction history into a savings plan, with human-in-the-loop review of AI classifications and an OpenAI Agents SDK workflow.
Experience
Co-founder & CEO, Alpha Flow
Building world-model training infrastructure for AI-driven quantitative trading. Our research includes MicroWorld, a multi-agent world model of US equity markets (with co-founder HongJin He).
Head of Events, NUS Chinese Debate
Run orientation and year-round recruitment, international friendly competitions, and overseas debate trips.
Education
National University of Singapore
GPA 4.8 / 5.0
Stanford University
- CS 229 Machine Learning
- EE 364A Convex Optimization I
- ME 344S HPC-AI Summer Seminar Series
Honors
- Science & Technology Scholarship, National University of Singapore
- Champion, Singapore Chinese Debate Championship