Hi, I’m Phát (you can call me aquaholic too) — a final-year Computer Science student at the University of Engineering and Technology, Vietnam National University (VNU-UET), and a Machine Learning Engineer at MoMo, Vietnam’s largest e-wallet platform.

I write about whatever I’m currently building or thinking about. Most posts here are things I wish existed when I was figuring them out myself.

---

Work

At MoMo I work on ML systems in production — recommendation, risk, and the data pipelines that keep them running. The culture there is deeply data-driven: every decision starts from evidence, every hypothesis gets tested. That habit carries into how I do research.

---

Research

My BSc thesis (2026) is on causal discovery — specifically, classifying the causal role of variables relative to a known treatment-outcome edge, using deep learning on observational tabular data. The work is grounded in the ADIA Lab Causal Discovery Challenge on CrunchDAO.

Starting from a reimplementation of the competition’s top-1 solution, I developed a series of improvements — multi-bandwidth kernel features, ANM residual channels, structural attention bias, and a 2D scatter density representation — that pushed balanced accuracy from 74% to 81%+, surpassing the original competition winner at 76.7%.

The full writeups are on this blog.

I had no background in causal inference before this. I picked it because it was hard and outside my comfort zone. I’d recommend that approach.

---

Interests

• Machine learning systems that have to work in the real world, not just on benchmarks
• Causal inference and what it means to understand data rather than just fit it
• Fintech and quantitative methods
• Occasionally: compilers, systems programming, and things that are closer to the metal than I usually get to work

---

Contact

• GitHub: aqu4holic
• Email: aquaholic1509@gmail.com
• Discord: aquaholic1337

Send me a message if you want to chat!