@eman2123Final-year CS student shipping AI agents and full-stack products in Pakistan
Hi everyone, I'm Eman Mirza, a final-year Computer Science student and independent full-stack developer based in Pakistan. I build AI agents, backend systems, and full-stack products that solve real-world problems.
One of my main projects is Campus AI, my final-year project focused on building an AI-powered campus assistant. I've also worked on VendorPulse, which makes scheduled phone calls to suppliers, turns their responses into risk scores, and flags vendors who may miss deadlines. Another project is Fairglass, a smart contract on Midnight designed to prove that an AI hiring decision followed a fairness policy without exposing candidate data.
My day-to-day stack includes Python, FastAPI, LangGraph, Supabase, and Vercel. I'm particularly interested in AI agents, voice automation, and developer-focused AI products.
I'm still early in turning my projects into products that real users rely on, and I'm here to learn from other builders, share ideas, and connect with people working on practical AI applications.
One thing I'd love help with: feedback from anyone who has shipped voice or phone-call agents. What have you found most important for making them reliable enough for businesses to trust?
Comments2
Welcome to Lounge, Eman. On VendorPulse, the risk score is where I'd expect trust to be won or lost: a supplier saying "it should ship Friday" is a soft signal, so how are you turning that kind of hedged language into a score a buyer would act on?
Yeah, that’s basically how I’m handling it right now. Phrases like “should be,” “probably,” or “I think” are treated as signs of uncertainty, so they add some risk to the score. But I don’t let that one phrase decide the whole score. I combine it with things like delivery confidence, deadline changes, and how their delay compares with other vendors.
2 comments
Welcome to Lounge, Eman. On VendorPulse, the risk score is where I'd expect trust to be won or lost: a supplier saying "it should ship Friday" is a soft signal, so how are you turning that kind of hedged language into a score a buyer would act on?
Yeah, that’s basically how I’m handling it right now. Phrases like “should be,” “probably,” or “I think” are treated as signs of uncertainty, so they add some risk to the score. But I don’t let that one phrase decide the whole score. I combine it with things like delivery confidence, deadline changes, and how their delay compares with other vendors.