A field guide to my obsessions.
A working catalogue across AI, robotics, and systems. Some are shipped and serving users; others are still cooking quietly in the lab.
Filter by discipline. Open any entry for the full case study — the question that started it, what got built, and what it cost.
A 3-service career analytics platform. 10 ATS signal extractors, 9 analytical lenses, hand-rolled SVG charts, and a runtime mock/live switch — deployed to AWS ECS with 213 passing tests.
002An RL-trained autonomous delivery robot navigating a 22-room Gazebo simulation. SARSA Q-learning over 30,000 epochs — learns to prioritise urgent rooms without being told to.
003MSc dissertation combining ASP logic planning, neural behaviour inference, and BREMEN offline RL in a single architecture — 30% lift over baselines, no shared policy required.
004A horizontally-scalable crawler: a stateless Flask/Gunicorn web tier and Redis/RQ worker tier sharing state in PostgreSQL, with a resilient fetch layer and CPU-based autoscaling on Kubernetes.
The shelf is not yet full.
A few projects live on private repos and behind NDA. If something specific would help your decision, ask — I keep a small library of write-ups for the curious.