Compliance paperwork, records chasing, lead handling, monitoring — one clear outcome, checked by a human before it reaches your customer, priced simply. No jargon, no strategy decks. Systems that hold up to inspection.
Paperwork drafted, records chased, filings prepared, incidents flagged — one metered outcome delivered every week, always reviewed by a human before it reaches your customer.
I connect the tools you already use — email, WhatsApp, spreadsheets, industry portals — so information moves itself instead of waiting for you at 9pm.
Control systems, embedded firmware, computer vision. For engineering firms that need someone who speaks both power systems and Python.
Things I've actually shipped. Commercial systems first, engineering builds after — with footage where it exists.
A digital CD register for vet practices: badge-login kiosk by the drugs cabinet, witness sign-off, running balances, discrepancy logging and audit-ready PDF export. Built to survive an inspector, not a demo.
Market-intelligence and outreach system for UK retrofit installers — finds grant-backed work in the EPC register and connects it to the firms equipped to do it. Customer dashboards, ranked feeds, automated compliance-checked outreach.
WATCH THE FOOTAGE
Built and programmed the SO-101 robotic arm for AI-driven manipulation tasks — footage of it working on the bench.
WATCH THE FOOTAGE
Built a working Bitcoin miner and work with the open-source Stratum V2 mining-protocol applications in Rust — real systems engineering, not crypto hype.
WATCH THE FOOTAGE
A smart bin that identifies and sorts waste automatically — sensing, actuation and product design, working on camera.
University team project: trained a YOLOv11 vision model to detect floating obstacles — wildlife, driftwood — so an uncrewed boat can navigate safely. Pictured in the test pool.
Ingests social-media posts in real time, has an LLM judge each one for market impact, and converts judgments into sized, stop-protected paper trades. 519 simulated trades, latency-decay and look-ahead-bias analysis — written up as an honest post-mortem.
Swing-up and stabilisation of an inherently unstable system — real-time control engineering, from mathematical model to hardware.
Autonomous maze-solving robot: embedded firmware, sensor fusion and path-finding on a hardware budget.
University project: a wearable tremor-measurement device — biomedical sensing and signal processing.
Fifteen minutes to find the one job that eats your time — the paperwork, the chasing, the monitoring. If I can't help, I'll say so on the call.
I run the job myself first, with AI doing the heavy lifting and me checking every output. You see finished work before any software exists.
What works gets automated and keeps running weekly. Simple price per outcome, cancel anytime, and I stay the engineer on the other end.
I'm completing an MEng in General Engineering, specialising in electrical engineering, and I've been building software the whole way through. That mix matters: the hard part of most automation isn't the AI, it's the real-world process it has to respect — a controlled-drugs audit, a grid-connection application, a grant evidence pack. I build systems that hold up to inspection, not demos.