Building AI products that run where people are.
I build software that puts AI to work for real people.
About
About
“I turn ideas into software that ships, and I care just as much about making it understood.”
Hi, I'm Juwon Lee, a product engineer who takes ML from dataset to deployed interface. I care most about the moment an idea becomes something people actually use, and understand. On most of my projects I've owned the work from planning through deployment.
My core is web and application development: Next.js and React on the front, with Supabase, Firebase and serverless functions behind them. I wire AI in as a working component rather than a demo, from an in-browser MobileNet v2 transfer-learning module students train themselves to an OpenAI image backend that watermarks and stores everything it generates.
Projects
Architecture deck (PDF)
Side & toy projects
Smaller experiments built along the way.
Tech Stack
AI & Computer Vision
I train, evaluate, and ship vision models end to end.
Data & Datasets
I build and check the datasets models learn from.
Frontend & App
I build the interfaces people actually touch, web and mobile.
Backend, Infra & Ops
I run the data, storage, and deploys behind them.
Hardware · IoT · Robotics
I connect sensors and machines to the web.
Planning & Education
I turn complex tech into experiences anyone can learn.
Work Experience
- Research, build, and run the interactive AI education web apps visitors use live on the exhibition floor; the Art Lab ran there through June and July 2026, going from 3,049 images in June to over 8,000 in July.
- Design education programs end to end: a smart-farm curriculum commissioned by Yangpyeong Education Office, built with its full server and web-app stack, taught as a demo class to 5 high-school students in August 2026, with the 15-student cohort class scheduled for October 2026.
- Ship the internal ops tooling behind them, such as the staff scheduler.
- Guide exhibitions bilingually (KO / EN) and support the multipurpose education rooms.
- Run the Picabot robot-arm drawing studio, a regular 30-minute hands-on AI session for visiting school groups; led its extended 50-minute class for the 100-student KT × Seoul City AI Future Camp in August 2026, co-hosted with KT, the Seoul Metropolitan Government, and Seoul Dobong Police Station.
- Built and reviewed visual question-answering (VQA) datasets for LLM training on an ETRI (Electronics and Telecommunications Research Institute) project, including complex 3-hop reasoning Q&A sets.
- Checked each question–response pair against the project's correction guidelines for grammatical and semantic accuracy, and rewrote the assistant answers that did not follow logically from the user's question.
- Sustained about 2.5× the team's average throughput while holding accuracy.
- Redesigned the national Pay-TV survey and drafted Broadcasting Act review materials.
- Covered 11 ministry exhibitions and conferences; authored reports and issue briefs.
Activities
Awards
GH Youth Build-Up Start-up Competition
PM & Developer
YAKMOA, a multimodal-AI medication-management service for digitally vulnerable users, built on the ATO care robot.
Science Museum & Community AI Hackathon
PM & Developer
An AI docent concept for museum-community co-prosperity, connecting exhibitions with the local community.
Chungnam Generative AI Start-up Idea Competition
PM & Developer
The 2024 VGG16 version of Doctor-Green (a plant-disease diagnosis app), pitched as a generative/AI start-up idea, hosted by the Ministry of Science and ICT.
HKNU StarUP&GO Audition
PM & Developer
Pitched the plant-disease diagnosis app that later grew into Doctor-Green, at Hankyong National University's start-up audition.
ICT·SW Women's Start-up Competition
PM & Developer
A plant-disease diagnosis app powered by a VGG16 image classifier, the predecessor of Doctor-Green (Korea IT Businesswomen's Association).