About
Full-stack AI engineer, based in Lahore, Pakistan.
For the last few years I've built production software across medtech, edtech, real estate, and SaaS. Mostly RAG pipelines, multi-agent workflows, and voice agents.
I take products from the first idea all the way to deployment and own every layer: architecture, backend, frontend, AI pipelines, infra, and CI/CD. I'm at my best early on, when a team needs someone who can make the architecture calls, plan out the AI, and get it in front of real users. I once took a Replit prototype to production in under two months. It reached number 5 on Product Hunt, and I've scaled platforms to thousands of active users.
I build the modern way. I lean on agentic coding instead of writing every line by hand. I drive Claude Code with my own slash commands, sub-agents, and skills, so the research, setup, and repetitive work happen in parallel while I focus on the parts that need real judgment: the architecture, the trade-offs, and getting it right. This isn't vibe-coding. I read, test, and own every change. It's the same care I've always taken, just shipped a lot faster.
- Ship to production, not demos A prototype that works in a notebook isn't done. I care about the version real users depend on.
- Own it end to end From architecture to deployment. I'd rather understand the whole system than just one slice of it.
- Make the AI earn its place Agents and RAG where they really help, plain and boring code everywhere else.
- Fast, but it has to hold Speed matters early on, but not if the thing falls over the moment real users show up.
- Modern tools, senior discipline I use agentic coding to move fast, then hold every change to the same review and testing bar.