Experience

SoundPatrol, Dec 2024 - Oct 2026

I prototyped a novel distortion-free watermarking method for generative AI music, with 100% detection in high-entropy generations. I aligned 1700+ hours of MIDI-audio data for fine-tuning a transcription model, filtering out 95% of mismatched pairs. I also created new scanners for YouTube, TikTok, and SoundCloud, increasing relevant dashboard results by over 10x, and deployed a scalable audio tagging model for music/non-music separation with over 99% accuracy.

ChucK Development Team, Mar 2024 - Jun 2026

I'm developing new features for the music programming language ChucK, in collaboration with several PhD students at CCRMA and the creator of ChucK, Professor Ge Wang. I helped design SMucK, a system for simplified musical input and dynamic playback in ChucK. SMucK enables users to import MIDI/MusicXML files into ChucK, and manipulate their playback in real-time.

Wati, Jun 2023 - Aug 2023

I trained several ML classifiers to predict if contacts would become customers of Wati, after setting up classes to collect and preprocess a changing dataset of contacts. I achieved 98% training accuracy and 92% test accuracy with initial models using a class-balanced dataset. I also constructed a data pipeline in Python to automatically query the YouTrack API and upload ticket information to BigQuery. I then visualized the progress of tickets over time and employee performance with Looker Studio.

Stanford ACM, Oct 2021 - May 2023

I worked with a team to build a convolutional neural network using Python and PyTorch to predict average income based on satellite imagery. We achieved the 3rd best validation loss out of Stanford teams who completed the project. I also mentored teams working on a bird classification model and an NLP model to identify human values behind arguments.

Stanford University School of Engineering, Mar 2022 - Dec 2022

I instructed moderately sized groups of computer science students in Python and C++ in the CS106 program. I graded assignments/exams and provided feedback to students. I worked closely with professors and other course staff to improve class material.

AVEVA, Jun 2019 - Jul 2019

I led a project to create a real-time, weight-based inventory management system using company software. Our team constructed weight sensors and wrote Python code on a Raspberry Pi to collect and upload weight data. We created an online, real-time visualization of the quantity of snacks in kitchenette snack bins as a proof of concept, shown in the GIF above.

AVEVA, Jun 2018 - Jul 2018

I modified Python code of 4 retro video games to receive real-time data to influence game mechanics. We used a REST API to relay data from the company data management system into the retro games. I also wrote a Python script to automate weekly time sheet entries for interns.