Agentic AI, engineered for production.
I'm Muhammad Abdullah, an AI engineer in Lahore. I build agentic systems, evaluation harnesses and the backend that runs them - for teams that need more than a demo.
Seven years, one direction.
From backend engineering for healthcare and finance to voice, recruitment and agentic AI, and production LLM work for Google.
Download résumé →AI Engineer · Turing, for Google
REMOTE, UNITED STATESAI Engineer · Nexxt.ai
REMOTE, UNITED STATESMachine Learning Engineer · Ascendia AI
REMOTE, GLOBALSoftware Engineer · Infostack
PAKISTANBS Computer Science · University of the Punjab
Lahore, Pakistan.
Four layers of AI, one engineer.
What I build, end to end.
The concrete systems behind each area of expertise - the things you can hire me to ship.
Case studies, start to ship.
Each one opens its own page: the problem, the system I built, and what changed.
From job board to a scored interview report.
AI engineer on a recruiting platform - candidate matching, outbound calling, a sub-500ms voice interviewer and scored reports, end to end.
PSX Equity Intelligence Engine
Analyst agents report to a supervisor, and 18 MCP tools do the math behind every verdict.
Read →TURING · 2026AI slides in each teacher's voice
Sole engineer on TeacherAIde - style fingerprinting, a Claude deck pipeline and billing.
Read →STAMPA SOLUTIONSHealthcare & fintech backends
Led a team of five building backend and automation systems.
Read →Six steps. No black boxes.
Understand
Pin down the real problem, the users and what success looks like.
Architect
Design the system - models, data, jobs and APIs - before writing code.
Prototype
Prove the riskiest part first, so decisions rest on evidence.
Build & ship
Ship in small, working increments you can see and test.
Evaluate
Measure quality with real evals, then iterate on what the numbers say.
Scale & support
Monitor, harden and scale in production - and stay close for what comes next.