Who I am
I read BSc Computer Science (Artificial Intelligence) at Royal Holloway,
University of London, where I finished my first year on a 74% First Class
average with 94% in Mathematical Methods, and hold the ISC Excellence
Scholarship.
Since June 2026 I have been a funded Undergraduate Research Fellow at
RHUL's DICE Lab — one of eight selected from a field of thirty-plus —
working with Prof. Kostas Stathis and Dr. Agnieszka Mensfelt on LLM
autoformalization. The work asks a narrow question with wide implications:
when a model produces a correct formal specification, has it induced a
general rule, or has it enumerated the facts it was shown? Answering that
honestly turned out to be mostly a problem of evaluation design, which is
where most of my time goes.
Alongside research I am an early engineer at Sharper Future Ltd, shipping
production features across the stack with LLM APIs and evaluation tooling.
My contract carries an IP carve-out that keeps my university research
independent — a detail I insisted on, because the two should not be
entangled.
I am from Myanmar, and some of my work points home: MyanLearn is a Burmese
literacy platform on Cloudflare Workers AI now used in more than eighteen
countries. Separately, I am reimplementing Karpathy's micrograd from scratch
as groundwork for getting inside transformer internals properly rather than
treating them as a black box.
Education
BSc (Hons) Computer Science (Artificial Intelligence)
Royal Holloway, University of London
First Class
Sep 2025 – Jun 2028 · Egham, Surrey
- Year 1 average 74% (First Class); Mathematical Methods 94%.
- ISC Excellence Scholarship holder, Royal Holloway International Study Centre.
- Coursework: Mathematical Methods, Algorithms & Complexity, Machine
Fundamentals, Logic & Reasoning.
Research interests
Mechanistic interpretability and neurosymbolic AI: evaluating whether
language models form genuine abstractions or memorise surface patterns, and
how capability interacts with scaffolding to determine that. Adjacent
interests in knowledge representation, formal methods, multi-agent systems,
and the design of evaluation harnesses that resist flattering their subject.
Service & awards
-
Secretary
RHUL AI & Robotics Society
2025 – 26
-
Technology Officer
RHUL Neuroscience Society
2025 – 26
-
Student Representative
BSc Computer Science (AI) — elected for a cohort of 27
2026
-
ISC Excellence Scholarship
Royal Holloway International Study Centre
2025
-
Member of Mensa
Admitted 2024
2024