CV
Education
- B.S. in Computer Science, University of California, 2024
- Minor: Bioinformatics
- GPA: 3.947
- Honors: Dean’s Honor List (Winter ‘21 - Fall ‘23), Shivakumar Endowed Scholarship in Computer Science, Upsilon Pi Epsilon, Tau Beta Pi
Work experience
- Machine Learning Research Intern, Rahmani Lab UCLA (May 2023 - Present)
- First-authoring new contrastive optimization algorithm for improved generalization on out-of-distribution data.
- Implemented several variants of algorithm in TensorFlow available for use with any gradient-based model.
- Conducted several experiments with various gradient filtering criteria, aggregation methods and developed adaptive learning rate methods based on cosine similarities and subset sizes for improved generalization performance.
- Created and visualized several diagnostic metrics including Jaccard indices for monitoring subset stability and graph-visualizations of final gradient coefficients across domains to identify intrinsic biases in dataset across multiple points in loss landscape.
- Achieved 5% improvement on median AUC performance on CONVERGE voice-recording datasets and matched benchmarks on Camelyon-17-WILDS.
- Supervisor: Prof. Elior Rahmani
- Machine Learning Research Intern, Bouchard Lab UCLA (May 2022 - Present)
- Developed parallelized Jacobian backpropagation library achieving > 30x speedup over PyTorch and TensorFlow on ResNet-50 Jacobian backpropagation on NVIDIA T4 GPUs.
- Designed accelerated Levenberg-Marquardt method leveraging transformed-space inversion achieving over 4% test accuracy improvement over Adam on CIFAR-10 with identical architectures.
- Created stochastic Gram Gauss-Newton matrix construction method achieving 3x per-iteration speedup as well as improved generalization error over exact Levenberg-Marquardt method.
- Conducted extensive benchmarking on performance of several first-order and second-order optimizers on MSTAR and CIFAR datasets.
- Applying inversion algorithm to natural gradient and antenna design with method of moments problems.
- Supervisor: Prof. Louis Bouchard
- Flight Software Engineer, Unmanned Aerial Systems at UCLA (April 2022 - April 2023)
- Led development on “Ground Station” dashboard displaying live telemetry and map position from a remote drone and ground vehicle for AUVSI SUAS inter-collegiate unmanned drone competition.
- Implemented motion planning under curvature constraint algorithms for fixed-wing aircraft for AUVSI-SUAS Competition 2023.
- Wrote automated camera scripts to capture and process images for classification as part of competition objectives.
- Frontend Developer at ACM Teach LA (October 2021 - August 2022)
- Produced several tutorial pages and ‘Terminal’ React component simulating Linux CLI for learning lab used to train ~40 ACM Cyber members and event participants in Linux fundamentals annually.
- Designed and made Cartesian-coordinate exercises replicating Turtle movement for Teach LA’s Introduction to Python curriculum taught at a series of local schools.
- Student Researcher at Pioneer Academics (April 2019 - August 2019)
- Studied materials theory pertaining to strength, fracture mechanics, thermodynamics and quantum mechanics.
- Learnt molecular dynamics simulator packages LAMMPS and OVITO and ran atomistic simulations for various Aluminum alloys with Copper and Magnesium.
- Analyzed simulation data using Python and materials theory to evaluate tensile strength-fracture toughness tradeoff of Aluminum.
- Conveyed findings in a research paper graded 97/100 by Prof. Frank Peiris of Kenyon College and shortlisted for publication in the Pioneer Research Journal.
Activities
- Writing Assistant for Bioinformatics Textbook (April 2023 - June 2023)
- Wrote chapter on dimension reduction via PCA and t-SNE for use in single-cell methods for CS CM121 course taught by Prof. Harold Pimentel.
- Upsilon Pi Epsilon (Oct 2021 - Present)
- Tutor undergraduate students in Computer Science, Math and Physics for two hours every week.
Skills
- Programming Languages: Python, C++, C, JavaScript, TypeScript, Lisp, Shell, Ruby, SQL, Haskell, Prolog
- Machine Learning: PyTorch, TensorFlow, JAX, CUDA, Scikit-Learn, Matplotlib, Seaborn
- Data: PostgreSQL, Apache Spark, MongoDB, MapReduce
- Web Development: React, Node, Express, Mongoose, Jekyll, WebGL
- Software: Git, Docker, Linux, LaTeX
- Cloud: AWS