About Hanzalah Waheed

I am a software developer based in India, working on AI and applied AI products. Since August 2024 I have been a Software Development Engineer at StockInsights AI, where I build the AI layer of an equity research platform. That means a retrieval-augmented generation pipeline that automatically summarises more than 25,000 SEC and earnings filings a month, a multi-session AI chat system on top of it, and the frontend that analysts actually use.

What I work on

Most of my work sits where messy data meets unclear requirements. I like the problems that start out badly specified and end with a system somebody can rely on. Recent examples: I migrated our full-text search off MongoDB Atlas onto Postgres, cut our Vercel bill by roughly 80 percent by tracking down scraper traffic that nobody had attributed, and shipped the company's first recurring API revenue using Stripe metered billing.

How I work

I care about understanding systems rather than assembling them. The tagline on my homepage — "trying to understand how things work" — is the honest version of my method. I read source code and primary documentation before reaching for abstractions, I measure before optimising, and I write up what I learn so the next person does not have to repeat the investigation.

Open source and writing

I contribute to open source, mainly to Interledger's Rafiki and to DocsGPT, where my pull requests have covered accessibility fixes, render-performance work, and refactors that remove duplicated component logic. I also write long-form technical posts about things I have had to work out the hard way: why updating a row in Postgres costs more than people expect, how to eliminate async waterfalls in React, and how to bring a JavaScript bundle back under control.

Tools

Day to day I work in TypeScript and Python, with Next.js and React on the frontend, Postgres and pgvector for storage and retrieval, and AWS plus Docker for deployment. This site itself runs on TanStack Start and Cloudflare Workers with D1.