I turn models into systems people can rely on, and interfaces people enjoy using.

I'm Shaheer Aslam, an AI engineer in London. I hold an MSc in Applied Artificial Intelligence from London South Bank University, awarded with Distinction, and a BSc in Data Science from GIK Institute in Pakistan. My work sits at the seam between modelling and engineering, where a promising prototype has to become a system that is fast, explainable, and trustworthy.

Lately that has meant agent reliability. I built an evaluation harness showing that LLM judges misread unsafe agent behaviour by pattern matching on tool names, and a taxonomy guided judge that lifted F1 from 0.46 to 0.92. I've also shipped an explainable churn model with a live dashboard, an autonomous cold outreach agent, a shadow mode deployment system, and a real time crypto data platform with full observability. On top of the systems, I build the interfaces: 3D scenes in Spline, wired up and animated with React and Framer Motion. A system that works but feels lifeless doesn't get used.

That range comes from working in different worlds: a software engineering internship at Jazz, freelance full-stack and mobile work, and a foundation in data science and interactive design. I care about evaluation, latency, monitoring, and the quiet reliability work that decides whether AI actually ships, and about the interface being something people actually want to use.