
Under the Hood of Vector Databases
A deep dive into embeddings, approximate nearest neighbors, and the architecture of similarity search, covering algorithms like HNSW, IVF, ANN, etc.
Moyez Rabbani
Hi, I'm Moyez — a Software Engineer from Kolkata, India with a genuine curiosity for the deeper side of Computer Science. I don't just use tools, I want to understand how they work under the hood.
Over the past 2.5 years, I've built end-to-end software that works well and looks even better. I care about design and user experience as much as I care about the code behind it — good software should feel effortless to use.
When I'm not coding, you'll find me reading or at the gym. I like building things, staying sharp, and never sitting still for too long.
Featured writing first, then the newest notes on engineering, design, and building useful products.

A deep dive into embeddings, approximate nearest neighbors, and the architecture of similarity search, covering algorithms like HNSW, IVF, ANN, etc.

Building RAG for 1,000 docs is a weekend project. At 10 million, it's a whole different story. In this article we break down everything to help you design one.

A step-by-step walkthrough of transformer architecture that follows one sentence from raw text to a translated output — with the attention math worked out by hand, not hand-waved.
The languages, frameworks, and tools I use to take products from idea to production.
Ideas I return to—kept as a small archive of reminders, observations, and hard-won truths.
“Twenty years from now, the only people who will remember that you worked late are your kids.”
“Grow or die. Maintenance is a myth.”
“It’s far better to know why you failed than to be ignorant of why you succeeded.”
“The first principle is that you must not fool yourself—and you are the easiest person to fool.”
“Practice does not improve you. Corrected practice does.”
“You make mistakes; you’re not a mistake.”
Leave a message here, or send an email at moyezrabbani.work@gmail.com