A senior engineer working across software, security, and AI. I build production systems, and the autonomous tooling that defends them — surfacing real risk early and making the fix obvious.
My route here ran through circuit boards and signal processing: a first-class degree in Electrical & Electronics Engineering, then years shipping production software in TypeScript, React, Next.js, and Python. I'm as comfortable in a browser tab as I am close to the metal.
As a Senior Software Engineer in DevSecOps at the London Stock Exchange Group, I put AI to work on security: the tools and autonomous agents that help teams find, understand, and fix risk before it becomes a headline.
What holds my attention is craft — architecture that stays clean as it grows, feedback loops short enough to think in, and interfaces that make a complicated system feel like it's on your side.
3 roles · progression
Lead the engineering behind our AI-assisted security tooling, applying LLMs and autonomous agents across the software lifecycle.
Delivered full-stack features across product and platform teams, and helped turn secure delivery from a checklist into a habit.
Joined as a technology graduate and moved quickly from rotations into real engineering delivery.
2 roles · progression
Built broadcast-automation software that bridged software and hardware over serial, IP, and GPIO.
Commissioned, tested, and supported enterprise-scale broadcast deployments.
Autonomous agents, rigorously evaluated
Autonomous agents that take a security task from question to answer. A custom harness orchestrates their tool use, grounds them in real context, and grades every output for accuracy, safety, and reliability before anyone is asked to trust it.
Multimodal property search, automated
A property-hunting agent that reads listings the way a person would. It scrapes live data, uses multimodal models to weigh both the photos and the description, and scores each place against the criteria that actually matter. Full-stack, with REST integrations and a relational data model built to scale.
Real-time control, on silicon
Final-year dissertation: a soft-core microcontroller (MicroBlaze on a Spartan-7 FPGA) running as a quadcopter's flight controller. It fused IMU data, ran real-time PID stabilisation, and handled nRF24L01+ wireless control — all on bare metal.
Everything I self-host runs on a single ~$300 N100 NAS that sips about 10 watts: roughly 20 Docker containers, a full media pipeline, DNS-level ad-blocking, and remote access with zero open ports. Here's the full tour — and why every piece is where it is.
Autonomous security agents are easy to demo and hard to trust. The difference is an evaluation harness — and treating evals as a first-class part of the system, not an afterthought.
The fastest way to make security fail is to make it friction. Here's how I think about building tools developers actually want to use.
What designing a flight controller on bare metal taught me about building software for people — and why the constraints were the point.
City, University of London · Dissertation: an FPGA-based microcontroller for real-time control.
Whether it's a security tool that doesn't exist yet, a full-stack product, or a problem nobody's quite cracked — if it's interesting, my inbox is open.