Taeeun Kim
Atlanta, United States | Seoul, Korea
I'm a student developer and AI researcher who firmly believes that even the most advanced AI is meaningless without a platform or application that brings it to life for real users. I strive to bridge that gap, leveraging AI to build experiences that have tangible impact on the general public.
I'm currently a rising junior at Emory University, studying Mathematics and Computer Science. As Co-Hackathon Director and Project Lead at Project Emory, I organized Emory's first-ever CS Hackathon with an amazing team and led a student team to build a website for a local business. Now, as Vice President of Project Emory, I’m focused on building a community of curious, driven, and collaborative students, to support and encourage everyone on their journey.
In my free time, I enjoy kickboxing, singing and playing guitar, cafe hopping for a good cup of latte, crying to great movies, or eating good food with good people.
My Experience
Jan, 2025 - May, 2025
Undergraduate Researcher • MAIX Lab
- Advanced multimodal ECG representation learning by integrating concept-based dictionaries into clinical report text embeddings, improving signal–report cross modal alignment accuracy by 9% over MERL approach.
- Applied FAISS for scalable disease concept retrieval, reducing retrieval latency by 65% over lexical methods.
Jan, 2025 - May, 2025
Jan, 2025 - May, 2025
Software Development Intern • Lablup
- Contributed to Backend.AI FastTrack, a batch job workflow management service, by implementing multi-selection and bulk deletion for pipeline jobs.Improved UI/UX of FastTrack with features such as a ReactFlow quick delete for node edges, along with dockerized local workflows and collaborative code reviews.
- Developed and deployed PyCon 2025 booth event page, supporting 228 participants across 1,193 submissions.
Jan, 2025 - May, 2025
Aug, 2024 - Jan, 2025
Aug, 2024 - Jan, 2025
May, 2024 - Aug, 2024
What I've Worked On
Curaitor: AI-Powered Coursework Management Tool
A highly concurrent automated local file organizer and processor as Go backend, using the Actor Model with goroutines and channels, Gemini Go SDK, and mutex-protected caching with JSON-backed models for persistence. Expanded with Next.js, React, TypeScript frontend to display organized course materials, up-to-date quizzes and study guides, and interactive knowledge graphs across courses.
Rendezvous
Pinterest- and Reddit-inspired social platform built on a modular Django architecture with rich text editor, grid UI, authentication-aware navigation, and social features. Deployed on AWS EC2 using Docker, integrating Nginx, Gunicorn, and MariaDB with Docker secrets for secure configuration and production scalability. Currently down due to costs.
Good Hangul
On-premise Korean grammar correction platform powered by self-finetuned and open-sourced nllb-200-ko-gec-3.3B model hosted on Amazon S3, Flask API backend, PocketBase DB for collecting training data, and Next.js/React/TypeScript web interface. Repo under company assets. Corresponding Chrome Extension that automatically detects inputs and displays correction retrieved from koGEC API.