English / 한국어
Hi, I'm Jinsu. I build web apps and macOS tools. My recent work includes a trackpad launcher, speech coaching, and interfaces for reviewing video edits.
Portfolio / Resume / jinsu.build@gmail.com
I also contributed a merged fix to Apple’s coremltools: correcting layer_norm beta shape validation and updating the regression test.
Double-tap your thumb on the trackpad to open an app grid. I designed, built, and released this macOS app on my own.
The gesture detector uses the shape of the touch contact to distinguish a thumb from a fingertip. Calibration adjusts it for the user; the SwiftUI launcher provides search and app arrangement.
Swift SwiftUI Homebrew Open source, MIT
Gesture detection · Calibration · App grid · Install
A pig judge questions your purchases. You submit a photo, answer questions, and get a verdict to help you reflect on your spending.
We entered the hackathon as a team of five. I planned the product, set up the infrastructure, and wrote all of v1 and v2. Rules determine the verdict; an LLM writes the explanation.
FastAPI SQLite Amazon Bedrock EC2 / S3
Encouragement Award, Honam LLM Hackathon.
Verdict logic · Model integration · Project details
A speaking coach that connects feedback to the recording, so you can go back and hear the moment being discussed.
On this team of three, I built the STT/LLM integration and coaching engine. I also added report replay and MP4 export.
Speech APIs LLM integration TypeScript
Bronze Award, JBNU AI & SW Competition, software division.
Coaching engine · Replay and export · Project details
A video-editing tool developed with Jeonju MBC. It lets an editor review and adjust AI-suggested clips.
I handled the frontend on a team of four. I built the source preview, timeline, and in/out editing controls used to check those suggestions.
TypeScript Video interfaces Timeline editing
Excellence Award, Software Capstone Design Competition.
Review workbench · Timeline controls · Project details
The images are illustrations based on the projects’ interfaces and features. The source links show the implementation.
Frontend: React, Next.js, TypeScript, SwiftUI
Backend: Python, FastAPI, Node.js, SQLite
AI APIs and deployment: OpenAI, Gemini, Amazon Bedrock, EC2, S3, Docker
What I recently studied
From August to October 2026, I studied machine learning fundamentals, LLMs, and RAG at Furiosa AI Agent School. My practice code is in furiosa-practice.
I'm a Jeonbuk National University student, currently on leave.
I'm looking for an internship in AI application development or full-stack engineering.