Built with
What I do
I work across the whole stack from training models to shipping the interface around them so research-grade ideas don't get stuck in a notebook.
Featured work
'Sup
As CTO I owned the full stack Next.js, FastAPI and Supabase and led a team of eight developers shipping fast in a startup environment. It's challenging, high-velocity work, and watching the platform grow to 2,000+ users and get covered by The Straits Times has been the most rewarding thing I've built.
About me
I'm a 21-year-old developer who loves figuring out how things work. I'll be matriculating into NUS Business AI Systems, but my curiosity already takes me everywhere. From full-stack web development to the depths of reinforcement learning. I treat every project as a way to get sharper.
Achievements
2nd Place — IMDA Solid Pods Hackathon
2024
1st Place — TRAE MiniMax Hackathon
2026
3rd Place — PyCon SG 2026
2026
5th Place (University) — GovTech AI CTF, out of 500+ teams
2024
Honorable Mention — HackOMania
2025
Finalist — Build for Impact
2026
Capabilities
Experience
'Sup (AI EdTech startup)
Architected and maintained a high-performance Next.js and FastAPI platform, growing it to 2,000+ users and a feature on The Straits Times. Led a team of eight to build and maintain the product.
Dec 2024 — Jun 2026
Industry Platform
My first deep dive into professional software. I built a registration system from scratch, automated CRM tasks with cron jobs, and shipped an AI business-card extractor.
Mar 2024 — Jan 2025
Self-employed
Where it all started. I help clients bring ideas to the web — from initial design through to 24/7 hosting — learning to manage projects and deliver on time.
May 2022 — Present
Selected works
Where I explore new tech and turn ideas into things that run.
Pick a job and get one dashboard showing the roles you can reach, the skills slipping under you, and the single government-funded move worth making — built on Singapore's SkillsFuture Skills Framework dataset with a Next.js, FastAPI and Supabase stack.
A digital wardrobe app for cataloging clothes and planning outfits, built at the 2026 TRAE hackathon.
A clean, ad-free reading platform that extracts text from PDFs, Word docs and EPUBs, with natural-sounding AI text-to-speech via Kokoro and Piper.
A web extension and app that dynamically injects event maps into subreddits, with GitHub auth and collaborative chat.
Multiple competition finishes — three 3rd places and two 4th places — applying advanced ML to complex, real-world data challenges.
Explored decentralized data solutions built on Solid Pods, taking 2nd place in a fast, competitive build environment.
5th place (University category) in GovTech's AI Capture The Flag, out of 500+ teams — adversarial problem-solving at the intersection of AI and security.
A full-stack image classifier reaching 99.4% accuracy with a custom CNN trained across 15 vegetable classes.
A web app that predicts a car's price from its features, deployed end-to-end.
An in-depth site raising awareness of learning disabilities and how to support those who have them.
Generating images with several GAN architectures trained on the CIFAR-10 dataset.
Working through as many Gymnasium environments as possible to learn RL hands-on.
Comparing many network architectures to solve the Pendulum control environment.
Recognized through
I'm always open to new projects, hackathons, and collaborations. If you have an idea or a challenge, let's connect and make it happen.