I build systems that turn scattered, real-world data into decisions worth acting on. My final year project, ResQMap, took chaotic disaster reports and used a fine-tuned language model to surface the most critical incidents first. Before that, I was training retrieval pipelines to make medical information easier to trust, and interning at Nexium building AI-first web products.
I graduated from FAST-NUCES with a Silver Medal and I care more about whether something actually works under pressure than whether it looks good in a demo.
|
AI-Powered Disaster Response System Crowdsourced incident reporting, verified in real time, ranked by a DistilBERT model fine-tuned on CrisisMMD so the worst-hit areas surface first. 
|
Grounded AI Answers for Healthcare Retrieval-augmented generation over medical sources using LangChain and FAISS, built to reduce hallucination in a domain where being wrong has real cost. 
|
