Working with the Sponsored Products team to build machine learning pipelines for product recommendations on the detail page

Worked with the Game Dev team on a competitive multiplayer indie game and NFT launchpad platform

Worked with the Personalization team on Call of Duty's Message of the Day• Trained and deployed recommendation model responsible for sending personalized messages (store items, news, and game info) to over 10 million players on a daily basis• Improved click-through rate, player engagement, and annual revenue (+$XMM)• Performed weekly data analysis, presented findings to stakeholders, and designed new experiments

Interned with the Personalization team on Call of Duty's Message of the Day• Created an experiment service in Java for training, evaluating, and comparing recommendation models offline, leading to faster discovery of better performing models and the protection of player experiences from poor models• Implemented an unbiased, offline evaluation algorithm for accurately evaluating contextual bandit models 14x faster than online A/B testing using sampled logs from the online model (2 days vs. 2 weeks)• Parallelized evaluation algorithm to run models on 1000+ seeds through multithreading, which lowered the variance of results by 16%• Improved eCTR of models by automatically testing new reward functions, features, and hyperparameters

Worked with Professor Luisa Caldas and the Virtual Bauer Wurster team at the XR Lab @ Berkeley• Launched Virtual Bauer Wurster (VBV), an online multiplayer experience designed for architecture students to collaborate and communicate with others inside a virtual replica of Bauer Wurster Hall• Led development of an online model editor for students to efficiently create, edit, save, and upload 3D architecture models to VBV, allowing students to enter their virtual creation with others and interact• Networked diverse and interactive communication methods such as voice chat, text chat, images, videos, post-it notes, laser pointers, and Slack

Worked with the University Health Service IT team• Provided front-line, technical support to all staff by responding to and resolving service requests • Imaged and deployed Windows OS to computers at the Tang Center• Installed hardware, debugged software, and solved networking issues

Worked with the GPU Compute Architecture team to build highly efficient GPU architectures• Developed Python scripts and connected with existing machine learning frameworks to automate machine learning workload validation and performance profiling on in-house GPU kernels• Generated data visualizations of test results to perform workload analysis and identify performance bottlenecks• Integrated multithreading and fault-tolerant features to optimize for automated, week-long jobs

Worked with a team of Oakland Technical High School students• Assembled a smart trash can that automatically sorts waste into recycling, composting, and trash• Programmed Raspberry Pis in Python and OpenCV to operate motors and recognize objects• Utilized Intel's Movidius Neural Compute Stick for computer vision inferencing

Interned with the University Health Services, Info Security, and Research IT team• Imaged and deployed Windows OS to computers at the Tang Center• Set up virtual machines and used Kali Linux for ethical hacking• Created a tutorial on using Docker for research students

Worked with a team of Oakland Technical High School students• Engineered a self-watering device for plants to address Oakland’s problem of food deserts• Programmed Arduinos in C++ to monitor moisture levels and supply water• Built a controllable rover with automatic collision avoidance

Interned with the Asian Branch Library team• Used Cantonese to teach patrons at the Asian Branch Library digital literacy skills twice a week• Created a website in Chinese that features custom tutorials and helpful links for using the web

Interned with the CETF team • Connected with OTX West and the Chicana Latina Foundation to schedule and survey 40+ people through telephone interviews and an online survey to better understand our audience• Produced a report analyzing the Digital Divide and recommending solutions for accelerating the deployment and adoption of broadband

Worked with a cohort of Bay Area Youth• Interviewed residents, businesses, and nonprofit organizations to present a logic study on Oakland and its pressing issues• Organized a dialogue on police accountability by inviting the ACLU, Berkeley Police Department, and Berkeley Copwatch to deliver workshops and speak with the local community, strengthening relations between the local community and the police