Chamroeun Hongleng · Phnom Penh, Cambodia
I build and ship software — web systems, decision engines, bilingual education and agritech tools — in TypeScript and Python, and fine-tune Khmer speech models when a product needs one, with honest metrics and an evidence label on every claim.
Actively looking for a software engineering internship — full-stack, backend, or data and ML tooling — remote or Phnom Penh.
I build software end to end: web systems, auditable decision tools for smallholder agriculture, and bilingual education tooling — and I add the applied-ML layer when a product needs one, like Khmer speech recognition where public data is scarce. The work stays grounded in field research with real farmers and in the business questions these systems ship into.
Dual-degree student — Computer Science at Fort Hays State and IT Management at AUPP — and national runner-up in mathematics, 2025. Building CHNAI LAB with one teammate in Phnom Penh, Cambodia.
Four pillars, one direction
01
How intelligent systems are actually built: data, training, evaluation, and the honest limits of what a model can do — practiced on low-resource Khmer problems where shortcuts fail fast.
02
Turning models and ideas into systems people can use: applications, workflows, APIs, architecture, and the deployment discipline that keeps them dependable.
03
Whether a system is worth building: user needs, unit economics, operations, and the market reality that decides if something useful is also commercially viable.
04
What a system is allowed — and ought — to do: contracts, terms, policies, privacy, risk, and the human-approval points that keep consequential decisions accountable.
Now
The internship search is my active priority — software engineering roles: full-stack, backend, or data and ML tooling.
Growing CHNAI LAB with one teammate — building the studio's products and the standards we hold them to.
Rebuilding this portfolio around four connected pillars, with evidence labels enforced by the build system itself.
Studying contracts, terms, and policy topics through international business coursework alongside the CS and ITM degrees.
Sharpening my English at IFL — academic writing, presentation, and the business-communication skills the English for International Business major is built on.
Published the Kaskor ASR fine-tuning pipeline and the Chomkar Decision Grid engine as public MIT-licensed repositories.
Preparing the next validation step for Chomkar OrderLoop: one recurring buyer pilot with documented commitments.
Selected work
AI & Machine Learning · Software & Product Systems
Khmer speech-to-text on a fine-tuned Whisper-small — best checkpoint 17.48% CER (self-reported) — a reproducible pipeline from raw audio to an installable CLI.
chamroeunhongleng.meThis site: a portfolio where every claim carries an evidence label, and the build refuses to publish work that is not yet honest. Public demo
PhsarOSAn operations platform for small Cambodian shops, cafés, and restaurants: point of sale, inventory, customers, and expenses on one screen. Public demo
Chomkar OrderLoopA buyer-first coordination concept: one recurring buyer order assembled from documented smallholder commitments before harvest. Pre-pilot
LMS — Bilingual Learning Management SystemA bilingual (English/Khmer) learning management system for course delivery, assessment, communication, and academic analytics. PrototypeLearning in public
Building from both ends: classic ML fundamentals in Python and C++ for depth, and fine-tuning real speech models for Khmer for practice — low-resource constraints punish cargo-cult ML quickly.
Learning to design systems before writing them, ship small typed and tested slices, and lead AI coding agents the way a reviewer leads a team — critically, with explicit approval gates.
Learning to test commercial assumptions in the field before writing code — and to tell the story afterward, through digital marketing and bilingual technical content.
Learning how contracts, terms, and policies decide what AI systems may do — using research agents to survey the primary literature, as a builder who reads the rules, not as a lawyer.
How I work
Code owns calculations, invariants, and permissions. AI helps research, explain, draft, and review — it does not invent the numbers.
Every important claim on this site carries a label: repository evidence, public evidence, owner confirmed, demo, planned, or private. Released, prototyped, and planned work are different claims.
Models can recommend; people remain accountable. Deployment, money, contracts, privacy, and public claims always pass through a human decision.
Low-resource Khmer data, field operations, connectivity, and user trust are product constraints — not edge cases to patch later.
Validate the riskiest assumption with the smallest working system before building the architecture the idea might one day deserve.
Everything here moves from idea to production through explicit human approval gates — the full process and AI policy live on the colophon.
Selected evidence
Awarded a 100% bachelor's-degree scholarship to the American University of Phnom Penh as second-place laureate in mathematics, on a rector-signed certificate dated 30 April 2025.
Direction
I would like to help build practical AI products for Cambodia: speech interfaces that work in Khmer, coordination tools that farmers and cooperatives can actually trust, and enough evaluation discipline to know whether any of it works. I am early in that, and the projects here are the steps so far.
The next steps are already on the roadmap: a model card for Kaskor ASR, one recurring-buyer pilot for Chomkar, and publishing evaluation results for Khmer speech.
I am a student looking for internships, research opportunities, and product work where the technical side meets business and governance questions. If that sounds useful, I would be glad to hear from you.