AI safety research
AI safety research
Researcher in the LASR Labs Summer 2026 cohort, where our team works on the science of model organisms. Papers on evaluation and on where machine learning actually helps.
Read moreAI Safety Researcher, LASR Labs (Summer 2026)
I work on AI safety evaluations and build trust infrastructure for AI agents
Founder of VibeTensor. Available from 19 Oct 2026, remote.

At a glance
What I do
AI safety research
Researcher in the LASR Labs Summer 2026 cohort, where our team works on the science of model organisms. Papers on evaluation and on where machine learning actually helps.
Read moreAgent identity and trust
Attestix, an open-source identity and attestation layer for AI agents: credentials, delegation and hash-chained audit evidence.
Read moreProducts and engineering
Founded VibeTensor, a DPIIT-recognised startup, and shipped Attestix, VibeMCP, IssueVista and UpdataMind.
Read moreCurrent work

Agent identity
ActiveOpen-source (Apache-2.0) core for AI agent identity, verifiable credentials, delegation chains and hash-chained audit evidence. 47 MCP tools and verifiers in Python, JS, Go, Rust, Java and R, producing compliance evidence for EU AI Act requirements.
Workshop paper
AcceptedWhat an interpercentile decomposition of multi-turn evaluation results can and cannot show. Accepted as a poster (non-archival) at the NeurIPS 2026 Trust-AI-Eval workshop.
Biosecurity
PrototypeOpen, vendor-neutral verifiable-identity and tamper-evident audit layer for biosecurity, sitting on top of DNA-synthesis sequence screening. Personal project, Apache-2.0.

AI control
HackathonControlArena setting for identity-delegation attacks, built at the AI Control hackathon in Mar 2026.

Evaluation tooling
LiveInteractive viewer that lets readers audit reported emergent-misalignment rates. Click any reported rate to read the generations behind it, with each judge's vote.

Open-source evaluations
MergedUpstream fixes to the open evaluation suite used by the UK AI Security Institute. A seeded answer-choice shuffle so every GPQA build is the same exam, and an IFEval install fix.
Experience
Jul 2026 - Oct 2026
LASR Labs
Funded place in the 13-week full-time LASR Labs research programme, supervised by Satvik Golechha (UK AI Security Institute).
Jun 2026 - Jul 2026
Iliad
Iliad Research Intensive (Iliad x PIBBSS), 8 Jun to 3 Jul 2026; attended all 20 teaching days.
Oct 2025 - Jan 2026
METR
Study participant in a 60-participant randomised controlled trial on AI coding tools, invited because of my open-source work.
Jul 2025 - Oct 2025
Insight Labs AI
Full-time role owning the AI agent layer of a trading platform.
Writing

Tutorial27 Dec 202512 min
State vanishes on refresh. Users lose their preferences, their history, their progress. I learned how to make Svelte 5 apps remember with localStorage and runes.

Guide26 Dec 202518 min
My open source project was a mess of scattered issues and forgotten PRs. I implemented ISO 21502 and PMI PMBOK standards. The result surprised me.

Tutorial24 Dec 202512 min
TypeScript errors were hiding in my codebase for weeks. The CI passed every time. One character in my GitHub Actions workflow was the culprit.
Reach out
Research collaboration, AI safety evaluations, or agent identity and compliance tooling: send a line about what you are working on.