Research related to Spatial Genomics

Contributing to Esper Explore Software.

● Improved image processing algorithms, cutting Esper Explore's build time by 60% and reducing product delivery by 1-2 days.● Crafted an intuitive visual representation of preprocessed spatial transcriptomics data across diverse hyperparameters, streamlining experimentation and spatial data analysis.● Created a GUI application for Esper Explore, which visualizes preprocessed multi-dimensional medical data for spatial data analysis.

● Developed calibration software for a model of LG medical monitors that is used worldwide.● Enabled users to customize the calibration routine based on their preferences and user history.● Incorporated WPF into the software’s desktop UI, which led to an increase in overall development productivity and code reusability.● Refactored 40% of the legacy codebase to .NET 6, increasing the software’s maintainability and efficiency.

● Developed a machine learning (ML) server that outputs a patient’s predicted risk level for heart disease based on their medical data, aiding doctors in their medical decision-making.● Developed an ML pipeline that allows engineers to upload and update their own ML models and automatically train them in batches with incoming pre-processed data, increasing the productivity of the company’s ML development.● Created a test framework in Python that covered 80% of the ML codebase.

● Worked on the backend for the subscription module of ‘Lundi,’ a food delivery platform.● Optimized the app’s implementation of the Haversine distance algorithm, resulting in a 50% faster search function.(Application Download URL: https://apps.apple.com/kr/app/lundi/id1501109408)

• Participated in a robotics research project that deals with helping blind people perform regular daily tasks with a robot.• Learned the basics of ROS as well as familiarizing myself with the research domain by reading over research papers.

● Supported the Optimal Intermittent Reorientation Project developed to create an AI model that imitates Scarabaeinae by applying deep reinforcement learning with trajectory data acquired from actual Scarabaeinae movement.● Contributed to tools to gather data that extract trajectories from videos assisting research in the deep reinforcement learning model (Python, OpenCV).

[Leading company in the electric scooter sharing service known as ‘Kickgoing’ in South Korea]● Efficiently managed and preprocessed a vast dataset of one million data points, integrating essential weather information.● Developed a user-friendly cluster map to visualize and analyze user usage patterns.