I started making classical music at 13 and learned by following the details obsessively. By 15 that had shifted into heavier and more experimental work, and by 17 I was producing trap metal and experimental music with a touring collective.
Atlas Reaper.
Systems engineer first, working across software, local AI, infrastructure, audio technology, and game development.
My route runs through music, aeronautical engineering, game development, and technical audio. Atlas Systems is the current live record of how those disciplines became one engineering practice.
Signal path
At 19 I went to Glasgow to study Aeronautical Engineering. COVID hit almost immediately and somewhere in that period of everything stopping, I realised I was studying the wrong thing. I left, worked full time for a while, kept making music, and started paying attention to what actually held my interest; which turned out to be systems. How things are built, how they connect, how they scale.
That led me to game design at Abertay University with a focus on audio systems, which clicked immediately. It sits at the intersection of everything I'd been doing without realising it: technical precision, creative problem solving, and building modular systems that behave predictably under pressure. From there I got into AI infrastructure, local LLMs, RAG pipelines, and the DevOps layer that ties it all together.
I'm currently in my final year, a recipient of the Saltire Scholarship, and building Atlas Systems as a live record of all of it — the projects, the decisions, and everything in between.
What I am concentrating on now.
The common thread is operational clarity: systems should expose their ownership, evidence, failure modes, and deployment state.
Automation and governance
Control-plane contracts, bounded automation, deployment evidence, and repeatable repository operations.
Local AI and grounded retrieval
Private model serving, memory, retrieval, and evaluation before a model or prompt is promoted.
Observability and evidence
Interfaces that keep declaration, observation, health, and publication as separate facts.
Portfolio-grade infrastructure
A public estate that is readable, independently deployable, and honest about what is implemented versus live.
Interactive audio systems
Real-time DSP, generative instruments, and sonification that expose technical behaviour through sound.
The rules behind the work.
- Evidence before deployment claims
- Explicit ownership and bounded permissions
- Fail closed when authority is uncertain
- Local-first AI and privacy where practical
- Evaluation before model promotion
- Automate repetition, not judgement
Working domains
A compact view of the four domains, ordered by the role they play in my current engineering identity.