About
How I ended up working across four fields
The story behind the projects — from mathematics competitions in Kampong Cham to speech models for Khmer.
From mathematics to computer science
I grew up in Kampong Cham and started with mathematics competitions: first in my province in Grade 9, national runner-up in Grade 12, and four full university scholarships when I left school. They taught me the habit this site is built on: an answer only counts when you can show that it is right.
I use the same habit in my projects. I work on Khmer speech recognition because the language I grew up speaking barely exists in everyday tools, and on agritech because the bok choy farmers our team interviewed in Kang Meas plant their fields without knowing who will buy the harvest. In both, careful testing matters more than an impressive demo.
Dual-degree student — Computer Science at Fort Hays State and IT Management at AUPP — and national runner-up in mathematics, 2025. Learning to work as an AI-native developer, and building at CHNAI LAB, a six-member student studio in Phnom Penh, where I work most closely with one teammate.
Why these interests connect
These are not four separate interests. The CS and ITM degrees build the ability to make intelligent systems. The international business studies and field research add the commercial view. The contracts, terms, and policy work asks what those systems are allowed to do, and what they ought to do. Every project on this site touches at least two pillars.
In practice, each project gets three questions instead of one: does the model work, would anyone pay for it, and what do the contracts, terms, and policies allow it to do? The projects here are student-scale attempts to answer all three, labeled honestly.
What I can contribute
- Shipping working software — TypeScript and Python end to end: a schema-validated Nuxt site whose build fails on unproven claims, a decision engine with 62 unit tests and CI, and a bilingual LMS prototype.Repository evidence: opens link
- Mathematical grounding — A decade of competition mathematics ending as national runner-up; the habit of proving an answer right before claiming it.Owner confirmed: opens link
- Applied-ML practice with honest evaluation — A public Whisper fine-tuning pipeline for Khmer with published weights, speaker-stratified splits, and self-reported metrics labeled as exactly that.Repository evidence: opens link
- Field research before code — Interviews with real farmers before writing anything, and a decision engine where refusal is a tested, first-class output.Owner confirmed: opens link
- Organizing work around written standards — I lead a small professional reading circle that runs on a handbook I wrote: rotating officer roles, progress measured by what members produce rather than by pages read, and an AI-use policy binding on me as much as on every member.Document evidence: opens link
- Bilingual delivery — Products, reports, and technical content that work in Khmer and English from the first draft, not as a translation pass.Public evidence: opens link
Skills
- Software — TypeScript, Vue/Nuxt, Next.js, Python, unit testing and CI, schema-validated content architectures, and AI-native development: leading coding agents with explicit human review gates. Current tooling: GitHub, GitLab, and AWS.
- Machine learning — PyTorch, Hugging Face Transformers, Whisper fine-tuning, CER/WER evaluation design, dataset and manifest discipline; currently studying classic ML fundamentals and C++.
- Product & business — structured field research, buyer-first validation, bilingual English/Khmer product design, digital marketing and short-form technical video (CapCut).
- Languages — Khmer (native), English (professional working).
How I work with AI
I work AI-natively: AI tools (primarily Claude) support my research, drafting, and implementation, while decisions, evidence labels, and anything that ships stay under human review — mine.
The full policy — what AI assists with and what stays human — lives on the colophon page.