About

I specialize in LLM orchestration, RAG pipelines, and NLP and care about evaluation, reproducibility, and making things work in production.

I built and actively develop ai-test-gen, a hybrid rule-based + LLM pipeline that generates structured test cases from user stories. Evaluated on a production CAD application, it delivers a 92% reduction in authoring time, 94.4% acceptance criteria coverage, and a 72.9% first-pass structural quality rate at ~$0.002 per test case (870 test cases for $1.74 of LLM cost). This is a real, production-grade system I continue to evolve and improve.

I'm currently completing an MSc in Computer Science (AI) at Queen Mary University of London, where my thesis research formalizes the methodology behind ai-test-gen.

Before dedicating myself to AI full-time, I spent years building large-scale engineering automation in enterprise environments. That experience shaped how I build AI today: clean interfaces, measurable outcomes, robust pipelines, and a strong bias toward systems that work reliably outside the demo.

On this site (and on Medium), I publish daily learning notes, Jupyter notebooks, and deep dives on ML, NLP, and deep learning, designed as fast refreshers for interviews and practical references for building real systems.

Open to: AI/ML Engineer · Applied AI · LLM / RAG Engineer · ML Systems / MLOps (London / Remote)