I've been shipping production software since the '90s — PHP sites on dedicated servers that lived in closets, back before AWS existed — and I've been making database calls for real money ever since. Over 25+ years I've helped everyone from solo founders building their first MVP to teams managing $10MM+ cloud budgets.
I've run a commercial product on a single SQLite file. I've wired a RAG pipeline of a dozen models onto an ordinary async stack. I've built the data pipeline behind hand-rolled neural nets in C at a wearable company, and the detection store behind a parking-garage license-plate reader where the database choices mattered as much as the model.
Some of those systems lost real data. All of them taught me something a tutorial never would.
Most database advice clusters at two extremes, and neither one is aimed at a small team. On one end, vendor content written to make you think you need a specific product. On the other, war stories from teams running thousands of nodes, solving problems you will not have at your scale. The middle — where a five-person team is trying to make one good call — gets almost nothing.
Meanwhile everyone's being told the same thing: you need a graph database for recommendations, a time-series database for metrics, a vector database for search, obviously, because it's an AI feature now. Follow that advice to its end and you're running five database systems with three people, two of whom also write the application code.
So I wrote the guide I wish I could hand to every founder I work with. Opinionated on purpose. It tells you to start with a boring relational database and stay there longer than the internet says you should — and it tells you exactly when that stops being the right call.
25+ years designing data layers that don't wake you up at 3am — from single-file SQLite products to Postgres with pgvector to managed cloud databases at scale.
Hand-rolled RNNs in C, RAG pipelines feeding a dozen models through LiteLLM, and the data pipelines underneath them. An ML system is mostly a fancy data system.
Black SEO Analyzer, a commercial product whose entire data layer is one SQLite file. Standing proof you can ship something people pay for on a single file.
Managed 70-person engineering teams and $10MM+ cloud budgets. Advised dozens of founders on technical strategy and the calls that actually move the bill.
Written for the team in the middle. Not platform engineers with a database team, not first-time CRUD builders. Small teams shipping AI features onto a product that already works.
Real systems, not whiteboard design. Every example comes from something I actually built or had to keep alive. No "imagine you're building a social media app."
Every chapter ends in a decision, not a menu. Opinionated defaults, and concrete thresholds for when to change course — not a comparison matrix you have to interpret yourself.
Honest about AI, including its own writing. The final chapter is a plain disclosure of exactly how this book was made.
Have questions about the book? Want to share your architecture? I'd love to hear from you.
Get the first 3 chapters free and start on the short path today.
Get First 3 Chapters Free