Jason Oh jasonoh1998

Hello, I'm Jason Oh (Cheolmin Oh) 👋

💼 Worked at:

Spare-it as a Data Scientist, where I built an AWS data pipeline to train segmentation models using YOLO. I leveraged SageMaker and its estimator library to pass Docker files for training with S3 buckets. To optimize costs, I developed a more efficient data pipeline using EC2 instances and S3 buckets, mounting them with the mountpoint-s3 library. I utilized custom IAM roles and AWS CLI API to replicate SageMaker's machine learning training process. I chose YOLOv8 for its versatility and open-source popularity, enabling me to deliver the most accurate segmentation models for the company.
BU Spark! as a Software Developer Intern. I led the development of end-to-end web and mobile applications using frameworks like Next.js, Vite.js, and Express.js, and databases like PostgreSQL and Firebase. My main role involved developing middlewares and using ZOD and TypeScript to ensure safety checks on API calls. Additionally, I was involved in UI development. As the software development lead on most projects, I interacted with clients, led sprints, and collaborated with UX/UI developers to build customized applications tailored to client needs.
South Korean Air Force Information Systems Management Group where I played a pivotal role in database management and managed internet and intranet network capabilities within the base. As a team lead for a year, I allocated tasks to team members, handled data preparation, and managed documentation for high-level officials.

🎓 Computer Scinece Major at Boston University, graduating in May 2024.

Spare-it: In the Spare-it project, my task was to build an image classification model to identify recyclables for Spare-it company. Link to our repository.

MyGenAI: Hosted with GitHub Actions and deployed to Azure Web App from a private repository. The description is available on the page. Mainly used LangChain and Streamlit.

Audio CNN Project: Specializing in audio data analysis, I focused on the application of spectrograms, mel-spectrograms, and MFCCs for gender and age detection. This initiative achieved high accuracy in gender identification and continues to improve age estimation.
BadLandLords: As the full-stack developer, I crafted scripts for automatic database updates through web scraping and APIs, utilizing TypeScript, Next.js, Prisma, GitHub Actions, Vercel, PostgreSQL, and Docker. Collaborating closely with a dedicated design team, we translated requirements from the Boston Council into a functional and user-friendly product, enhancing tenant access to information about responsible landlords. My role was pivotal in ensuring the seamless integration of design and development, aligning closely with the project's goals.

Shine-AI: Focused on mental health support, I led the development of an application using Next.js, Prisma, Firebase, Vercel, and Jest. My role involved designing backend services and integrating external support mechanisms to create a direct pathway for users seeking professional help.
D7: In this project aimed at enhancing community services in Boston's District 7, I spearheaded the development using Vite.js, Strapi, Firebase, Netlify, and GitHub Actions. My contributions included extensive refactoring for efficiency improvements and the integration of new features, alongside accurate API implementations. I worked closely with Boston City Councilor Tania Fernandes Anderson, ensuring the project's alignment with community needs.
Swan Stream: For this streaming service, I utilized JSP, Bootstrap, jQuery, Java, Spring, MyBatis, JSTL, and OracleDB, deploying it on AWS to create a robust platform for drama and comics.

💡 Proficient In (Everyday Use):

🏗 Intermediate (Know How to Use):

🌱 Basic Knowledge (Learning & Exploring)