TP
Thuc Phan
Login
/about

Building systems,
learning in public.

I'm Thuc Phan. I spend my days building data and AI systems, and my evenings writing notes, testing ideas, and building small tools to better understand the world.

discipline
Data & AI Engineering
domain
Fintech · Markets
writing
since 2026
building
3 live tools

What I do.

At Timo I lead the data engineering work — the whole arc from raw data to applied. Platform and pipelines first, then the governance that keeps them trustworthy, the analytics that turn movement into meaning, and the models and agents that act on it. I lead a small team; we own anything that touches transaction data or learns from it. The work I'm proudest of is rarely the model.

What I write about.

A handful of rough buckets — AI, data infrastructure, fintech, markets, and the occasional note on tools & craft. The connective tissue is something like: how do we build software that compounds without breaking the things underneath it.

The notebook itself is new — I started it in 2026 and the first issue hasn't gone out yet. The intent is one essay every other Sunday, with shorter notes in between when something's worth saying. No SEO, no growth experiments — just the writing.

What I'm doing now.

A live snapshot sits up top — where I am and what I’m focused on. The things I actually ship live on the projects page: GenPulse for market intelligence, KraftCard for spaced repetition, and the Notebook this site runs on.

What I work across.

Data Platformsbig data, streaming, and real-time systems
Analytics & BImetrics, dashboards, and decision support
AI Systemsllms, rag, and agentic workflows
Platform Engineeringkubernetes, gitops, and infrastructure as code
Cloudaws & gcp

A short principle.

Most software problems are really just memory problems wearing different costumes — caches, indexes, schedulers, embeddings, even the human brain. Spend most of your career thinking about what to remember, when to forget, and how to make the difference cheap.

Write back.

If anything here was useful, contradictory, or quietly wrong — I'd love to hear about it. I read every email.