Learn. Question. Feel. — A Product Data Scientist's Field Notes

I write a notebook of my experience. This is to similar to tutorial.

What I'll be writing about

These are the topics I'm actively learning, questioning, and working through on the job:

What I Build:

• Experiment Methodology — Sequential A/B testing, CUPED, ghost bidding, stratified sampling. The methods that make experiments trustworthy — or quietly break them.
• Causal Inference — Synthetic Difference-in-Differences, Synthetic Control, and the broader toolkit for asking "did this actually cause that?" when a clean experiment isn't possible.
• Online Advertising — The concepts behind impressions, auctions, CTR, ROAS, and how ad serving actually works at the infrastructure level.
• Experiment Architecture — The structural decisions behind an experiment: how samples are assigned, how units are split, how you avoid the subtle mistakes that invalidate results.

What I Learn:

• ML Products — How machine learning systems actually work inside a product. Ranking, recommendation, real-time inference. Not just the model, but everything around it.
• Backend Endpoints for ML & Ads — The API layer that connects models and ad systems to real users. Latency, feature stores, serving pipelines.
• Feature Flags — How flags gate releases and experiments, and why they matter more than most analysts realize.
• GrowthBook — An open-source experimentation platform I'm exploring for feature flagging and A/B test management.

(Might Build):

• Experiment Journal — Why documenting experiments matters, and how to build a system that lets you actually learn from them over time.
• Text2SQL — How LLMs translate natural language into SQL, what works, and what breaks in production.

Why tutorial-style

I want to write a context that I can come back to it later and immediately recall why it matters, where it gets tricky, and what and how to do it.

If someone else finds it useful, great. But honestly, the first reader I'm writing for is future me.

Why now

The work I do sits at the intersection of data, product, and engineering — recommendations, search, advertising, experimentation. It's a lot to hold in my head. Writing is how I make sense of it.

So this is me starting to write it down.