01 / Curiosity
I came to AI from international finance.
Both fields ask the same question:
what is this actually worth to someone?
BSc International Finance, Honours
MSc Artificial Intelligence, UDIT
02 / Care
Before a model reads someone’s hiring assessment, I want the parts that could be used against them already gone.
↳GDPR Article 9 screening serviceSole engineer · design to production
A screening service for sensitive data in hiring assessments, built from design to production. The work starts before the model: deciding what it should receive in the first place.
03 / Rigor
I don’t trust a system because it sounds right. I trust it once I’ve measured it and can show you the number.
↳Five detectors. One measured choice.F1 benchmark · 0.786 vs 0.755
Five detectors benchmarked; the choice made on measured F1.
F1 SCORE · HIGHER IS BETTER
04 / Service
The best thing I build is usually the thing that quietly stops wasting my colleagues’ time.
↳Multi-modal MCP toolsHelping the frontend team find bugs faster
Tools built around the frontend team’s debugging work, bringing multiple forms of context into the process of finding bugs.
05 / What’s next
Systems that are safe to hand over.
Where the guardrails, the evaluation, and the reasoning stay legible to whoever comes next.
↳MCP governanceRemote Kubernetes · three layers of auth
MCP tools moved to remote Kubernetes behind three layers of authentication. A practical expression of what I want to build next.