Tom Morris

RAMP Studio is a masterplanning and architectural practice based in the New Forest. I worked closely with the team to design, build, and launch the site, and am actively responsible for its maintenance. It's live now, take a look.

My final-year dissertation: continuous-depth neural networks, written completely from scratch in Rust with no autodiff library. That meant deriving the adjoint sensitivity method by hand and constructing my own adaptive, bidirectional Runge-Kutta solver.

Using the framework, I trained experimental models across a variety of problem domains: orbital patterns of Earth and Mars, the initial COVID outbreak in Italy, and the VIX volatility index. The project was highly commended by the assessors of that year's Duncan Prize, awarded for the best Senior Honours project in Mathematics or Statistics.

At its core, a neural ODE is simply a learned vector field; the figure below shows one such field evolving live.

A self-hosted Docker homelab running on Ubuntu Server. It started as a way to learn infrastructure and networking fundamentals hands-on, but has turned into something I use daily. It's taught me reverse proxying, DNS and certificates, VPN routing, and the Linux upkeep that comes with running your own hardware.

I'm excited to be joining the Delivery team in September to work on the systems that route and dispatch orders in real time, predominantly in Rust. More once I'm there.

I worked on the Switch Abstraction Interface layer of Spectrum-X, NVIDIA's Ethernet-based AI networking platform for data centres. I shipped an end-to-end packet routing feature in C for a major client's new data centre build, and wrote the validation suite for it in Python, run against both simulated and real hardware.

I also partnered with the SONiC integration team on platform issues arising on unreleased switch hardware, tracking down and resolving concurrency bugs across the SAI implementation.

I initially joined for a summer internship, building the company website solo in React and Next.js, and then spent the rest of it working on the iOS app in Swift: UI, some backend, and a lot of bug fixing. This is where I first learned to put a project together end-to-end on my own.

That turned into a part-time role alongside my studies for the year that followed. I resumed my work on the iOS app, and then moved onto creating a much-needed backend. I built it on GCP using cloud functions written in TypeScript, unifying every app behind one shared API instead of per-app database access.

I read maths and computer science at St Andrews, graduating with a first-class BSc (Hons), a 93.5% average across the degree, and a place on the Dean's List every year.

Most of what I enjoy sits where the two overlap: problems where the maths is the hard part and the systems work makes it real, which is roughly what drew me to networking and real-time systems.

Outside of that: hitting the gym, improving at golf, the occasional volleyball game, and building computers I don't need.

Messages come straight to my inbox. Whether it's project work, questions about anything on this page, or something you think I'd find interesting, leave an email address and I'll get back to you.