About
I'm David, a Data Science and AI undergraduate at NTU building toward a career as a machine learning / AI engineer in fintech. I'm drawn to the harder end of applied ML — AI agents, LLMs, and retrieval-augmented systems — and to the discipline that fintech work demands, where a model isn't judged by how it performs in a notebook but by whether it survives contact with real transactions and real users.
Most recently, I spent a summer as a data analyst intern, where I shipped a recommendation engine to 1.6 million users and built LLM-powered tooling that let non-technical teams query live data using natural language without waiting on engineering. Outside of internships, I build things that force me to get the fundamentals right — forensic ML systems, RAG pipelines with real data isolation guarantees, pricing models validated with proper A/B frameworks.
What I'm building toward
AI Agents & LLM Systems
I'm most interested in systems that reason and act, especially agentic pipelines and RAG architectures.
Fraud Detection & Risk ML
Fintech data punishes sloppy ML, where leakage, non-stationarity, and noise that looks like signal are the default failure mode. I care about getting the unglamorous parts right: proper validation, honest backtests, and features that don't quietly cheat — the same discipline fraud and credit risk models demand when they're making real-time decisions on real money.
ML/AI Engineering for Payments
I want to work where rigorous data science meets production systems — building and shipping ML models for fraud detection, credit risk, and other payments infrastructure that has to hold up under real-world scale and adversarial pressure.
Outside of code
I love running and football, and compete in the 110m hurdles. Football was my first love before I pivoted to track, but I still play football from time to time and am an ardent supporter of Manchester City, for better or worse depending on the week.
