• Built AI to determine optimal timings of battery cycles and environmental conditions, to prolong their lifetime.------• This AI consistently generates timing profiles with (est.) 4% lower capacity degradation rates than the standard battery test.------• AI developed in Python, using PPO algorithm from stable-baselines3 library• Built robust data processing pipeline (in Python) which delivered key insights on battery models’ performance, via data visualization charts.------• Pipeline used machine learning (Linear Regression model) to forecast batteries’ expected lifetimes with 96% accuracy.------• Insights from pipeline are being used by Power Sources Group to characterize aging rates of various battery models.• Built webpages (in Python Django framework) for battery test data website, enabling users to upload, query, or delete battery test data records from SQL database. Webpages are projected to be used by ~50 staff.

• Project 2: Managed team (6 students) in development of Reinforcement-Learning AI to control simulated racing vehicle (in Python, using PPO algorithm from stable-baselines3 library). AI can autonomously drive racecar through various racetracks with low lap time.------• Currently refining simulation code so that AI can be utilized to control lab’s real-world autonomous vehicle• Project 1: Managed team (4 students) in development of adaptive cruise control system (in Matlab and Simulink). System consisted three integrated PID controllers. Presented project poster at URSCA Research Conference 2023.• Project 1: Built various Graph Machine Learning models (in C++, MATLAB) from scratch, in order to forecast traffic flow on LA streets (accuracy achieved: 91%). Key models used: Graph Temporal Neural Network, Graph Attention Network, LSTM

• Built product-testing automation pipeline for diabetes management product. Pipeline gathered, parsed, and analyzed data on Bluetooth signal strength between wireless devices, amidst varying levels of signal interference.------• Implemented Gaussian Regressor (in Python, using Gpy library) to probabilistically model devices’ signal strength.------• Pipeline’s generated dataset enabled team to characterize/ verify communication strength between interconnected devices.• Constructed two Cloud-hosted applications (using Microsoft tools PowerApps, Lists) which enabled Systems Engineering Team to collect and manage issue reports from hundreds of Medtronic product demo-users.

• Built computer vision algorithm to track cell nuclei in video feed and extract data on cell position/size changes------• Algorithm automated team’s process of recording data on interactions between Zika Virus and premature brain cells------• Algorithm written in Python. Libraries included: opencv, numpy, cellpose, scikit-image

• Built battery pack state-of-charge estimation software tool (in C), using Kalman Filter algorithm. Tool was incorporated on vehicle dashboard, enabling driver to visualize battery pack charge level.• Built battery pack state-of-charge simulation tools (in MATLAB and C). These enabled team to test my state-of-charge estimation tool, prior to deployment.

• Implemented and tuned various machine learning algorithms (in Python) to predict DOE supercomputer experiment runtimes (accuracy achieved: 94%). These helped PACE Team understand the spread in runtimes across various experiment types.------• Regression models applied (via Python libraries Sklearn, Keras): Deep Neural Network, XG Boost, Random Forest, KNN, SVM------• Dimensionality reduction techniques: PCA, Random Forest Feature Importance, pairwise correlation• Built automation pipeline (using Python, SQL) to generate data table of supercomputer experiments’ memory usage. Pipeline continuously updated team with charts of memory usage of ongoing supercomputer experiments.

• Managed team that created C/C++ programming tutorial guides which were used by 200+ students in various UC Davis engineering courses. Was granted official page on UC Davis Canvas website to upload guides.• Key topics covered: data structures, object-oriented programming, algorithms, recursion, memory management.

• Constructed fundamental components of navigation system (in Python) for organization’s autonomous CubeSat project. (This CubeSat was recently launched into orbit by NASA’s CSLI Program).----• Used Python, MATLAB, Excel, and vector/ matrix mathematics to design mathematical CubeSat position mapping system

This program taught students JavaScript, html, and CSS. Sponsored by Beyond the Bell organization, Arc company, and Reseda high school.

- Team taught underprivileged students art of problem solving (domains: probability, statistics, algebra, geometry).- Team ranked amongst top 6.6% of schools in 2019 International Purple Comet Math Competition.