Computer Science student at Heriot-Watt University in Edinburgh. I work close to the hardware: FPGAs, embedded displays and a home-built NAS.
Things I've built or am in the middle of building. Each one gets a write-up covering what it does, how it works and what went wrong along the way.
My honours dissertation. I'm measuring what compression techniques such as quantisation actually save when a neural network runs on an FPGA accelerator: how much power, how much speed, and how much accuracy it costs.
A small home server for storage and backups, built in a Jonsbo N4 case around a Ryzen 5 5600GT with two 4 TB IronWolf drives. Backup scripts and a health-check dashboard come next.
A 2-inch screen and status LEDs on an Arduino Nano ESP32 that show how the NAS is doing at a glance. An analogue needle gauge for power draw comes in version two.
Network-wide ad and tracker blocking for my home network, running on a spare Raspberry Pi 3.
This site. A circuit board you click through, served from a Raspberry Pi 5 on my desk, with the source public on GitHub.
Computer Science student at Heriot-Watt University, working towards hardware, embedded systems and energy-efficient machine learning.
Dissertation on evaluating compressed neural networks on FPGAs, with a focus on power draw. Modules include Language Processors, Data Structures and Algorithms, and Foundations of Computer Science.
Fast, accurate service in a busy themed cocktail bar, alongside full-time study.
Testing, repairing and preparing donated computers for reuse in schools in sub-Saharan Africa.
Build logs and notes on whatever I'm working on. A new post goes up when a project reaches a milestone.
Why the homepage is a circuit board, how it's drawn and how the zoom works.
The case, CPU and drives I picked, and why a Ryzen APU suits a storage box.
Notes from my dissertation literature review on compressing networks for FPGA accelerators.
Most portfolio sites are a list of links. I wanted mine to look like the work I want to do, so the homepage is a development board. The FPGA in the middle holds my projects, the memory holds my CV, the crystal keeps time for the blog, and the network ports are how you get in touch.
A photo of a real board would have been quicker, but it would belong to someone else, carry their branding and look soft on a high-resolution screen. This board is SVG, built from a short list of components and where each one sits. Every chip is its own element, so it can react when you point at it, and it stays sharp at any zoom.
The details are deliberate. The traces between the FPGA and the memory wiggle because real DDR traces are length-matched so signals arrive together. The FPGA boots from the small flash chip below it, which is why the status LEDs flicker when the page loads.
Clicking a component doesn't scale an image. The SVG's viewBox, the window onto the drawing, animates from the whole board down to the part you picked. Because the drawing is vector, it's redrawn at full resolution on every frame.
// Move the camera between two boxes
const w = w0 * Math.pow(w1 / w0, t); // zoom changes at an even rate
const cx = focus.x - offset * (1 - t) * w;
svg.setAttribute('viewBox', `${cx - w/2} ${cy - h/2} ${w} ${h}`);
Once the part fills the screen, the page grows out of it. Going back runs the same animation in reverse.
| Component | Page | Why |
|---|---|---|
| FPGA (U1) | Projects | It's the part that does the work |
| DDR4 memory (U2, U3) | CV | It stores history |
| Crystal (Y1) | Blog | It keeps time |
| USB-C and Ethernet (J1, J2) | Contact | They connect the board to the outside world |
| Power regulator (U5) | About | It powers everything else |
The site is a set of static files served from a Raspberry Pi 5 on my desk in Edinburgh, and the source is public on GitHub. The status LEDs on the board will eventually show that Pi's own health.
I'm Hamish, a Computer Science student at Heriot-Watt University in Edinburgh. I'm most interested in the point where software meets hardware: how code becomes circuits, and how to make it run faster on less power.
My dissertation looks at compressed neural networks on FPGAs, measuring what techniques like quantisation actually save in watts and what they cost in accuracy. Outside of uni I'm building a home NAS, a small desk display that reports its health, and this site, which runs on a Raspberry Pi 5.
I also work behind the bar at a cocktail bar in Edinburgh, which is good training for staying calm when everything happens at once.
Email is the quickest way to reach me. I'm happy to talk about graduate roles, FPGAs or home lab builds.