
teaching machines to learn.
Education
• Coursework: Graduate Machine Learning, Efficient ML
• Coursework: Data Structures, Algorithms, Computer Organization, Systems & Networks, Machine Learning, Computer Vision, Numerical Analysis, Automata & Complexity, Financial Modeling, Management of Financial Institutions, Fixed Income
• Honors: Faculty Honors with 3.98 GPA
Work Experience
• Designed and developed a full‑stack generative AI product using Kotlin & ReactJS for AI Test Kitchen (AITK).
• Designed, developed, and shipped user‑centric features to AITK’s MusicFX by developing new RPCs, and utilizing Boq and Spanner.
• Worked across teams by owning the idea, getting PM/UX approval, writing comprehensive design docs, implementing full‑stack features with minimal disturbance, launching new features to production, and presenting to product area, without compromising quality.
• Implemented a new RPC on the server side and created an automated workflow using Java that saves 100+ engineering hours every year.
• Integrated with existing tools using Sheets API in Java and Golang, reducing deployment time by a week every release.
• Wrote a design doc regarding the reporting workflow, went through code reviews, launched internally for ChromeOS, and presented work.
Writing
• Pruning, Lottery Ticket Hypothesis, Vision Transformer, LOTUS
Projects
• Developed a new GPS system in Arduino, worked on the Sensor Acquisition Board design with EAGLE, and redesigned the website.
• Placed 3rd overall, 2nd in cost, 3rd in acceleration, & 4th in design in the 2022 Formula SAE Electric against 50+ collegiate teams.
• Placed 1st in acceleration and skidpad, finalist in design, & 2nd in autocross in 2023. Holds the North American acceleration record.
• Founding club member and revamped much of the communication, increasing the club’s size by 125+ students in its first year.
• Working with the executive board to create a new curriculum, bring in sponsors (i.e. FTX & TradingView), & organize a large trading competition for club members.
• Built a convolutional neural network using over 950+ manually labeled video frames with a pipeline for pre‑processing and filtering.
• Improved the model from purely convolution (CNN) to a segmented one with YOLOv5, increasing accuracy to 96%.
• Designed and implemented an Android application for the makerspace staff, providing real‑time alerts to users of unsafe band saw usage.
Volunteering
• Managed and coordinated the Disease Detectives event at UGA and Yellow Jacket (GT) Invitationals, ensuring adherence to competition rules.
• Oversaw all aspects of the event at the Georgia State competition, including participant registration, scorekeeping, and resource allocation.
Awards
• Project: Using CV to Augment Makerspace Safety
• Teammate: Inhyeong Park