AI/ML Engineer working on computer vision. I train models, build LLM systems and data pipelines, and ship them to real clients.
I'm an Associate AI/ML Engineer and a Computer Science graduate of FAST-NUCES (Dean's List). I work at the point where models meet real clients: vision systems that inspect damaged vehicles, Claude-powered tools that read engineering documents, multi-agent RAG for finance teams, and the Airflow pipelines that feed them.
What I care about most is honest evaluation. A good demo isn't enough for me. I dig into where a model fails, whether that's shortcut learning, leaky metrics or the wrong checkpoint, fix it, and explain the result in plain language.
Away from work I build things for fun: Urdu NLP tools, a gesture-controlled Spotify player, and a Ghibli-inspired bookshelf. I've also led events for my university's biggest festival. 📸
Trained YOLO instance-segmentation models that find vehicle components and damage in photos, then audited DINOv2 classifiers to find out where they actually fail (shortcut learning, crop artefacts, checkpoint selection).
Built and onboarded ingestion pipelines for a data platform that serves prediction-market and country-risk datasets to customers and AI agents, including historical backfills and scheduling fixes.
An AI copilot that reads site-plan PDFs and checks them against a county submission checklist. Confidence gates send uncertain items to human review, and the output is a cited comment letter and an Excel issue report. I led the live client demo.
Claude-integrated workflows that extract data from client documents, run Excel calculations and generate reports, replacing slow manual processing.
A multi-agent retrieval system over a client's internal finance knowledge base, with specialised agents that retrieve, reason and answer with source citations.
Requirements calls, technical scoping, MVP proposals and findings reports. I turn model results into plain-language decisions for non-technical stakeholders.
Client names and proprietary details are withheld under NDA.
Dean's List (2 semesters) · Event Management Lead, DAIRA · Social Media Head, FAST Photography Society
I'm open to AI/ML engineering roles and interesting projects. The fastest way to reach me is email.