cv

BA Computer Science

University of California, Berkeley

Software Engineer Intern

NVIDIA Corporation

Undergraduate Researcher

Berkeley Speech and Computation Lab

• Developing deep learning models for acoustic data using PyTorch and using speech to understand how neural nets learn.
• Modeling language acquisition with GANs and VAEs to develop interpretable systems to process language in humanlike ways.
• Exploring the application of diffusion models in computational models for the implementation of speech technologies.
• Constructing and applying neural networks to sperm whale bioacoustics data to learn meaningful representations of underwater whale vocalizations as part of Project CETI, machine learning and linguistics towards decoding sperm whale communication.

Research Intern, Software Development

MIT Lincoln Laboratory

• Developed a low-cost, rapidly deployable transponder for radar validation, saving 98% of prior transponder field deployment costs.
• Engineered supporting transponder codebase written in C++ to interface with hardware components, allowing for accurate flight tracking and logging, microsecond GPS timekeeping, and full-duplex Rx/Tx capabilities.
• Applied graph neural networks through PyTorch for radar signal classification to increase accuracy from prior YOLO model.
• Assisted operations and testing of HF over-the-horizon radar array project through onsite Agile software development and hardware maintenance at White Sands Missile Range, NM.

Research Intern

NASA Goddard Space Flight Center

• Developed and tested custom TensorFlow and Keras applications on the DGX high-performance computing (HPC) cluster to improve parallel training performance of deep networks.
• Drafted a standard tutorial and report for migrating geoscience deep learning applications to GPU and multi-GPU-based environments, increasing the efficiency of distributed computing on the NASA Discover supercomputer and Prism clusters.
• Delivered a package containing four geoscience deep learning applications for testing distributed machine learning platforms.
• Presented ‘AI/ML/DL Benchmarks for Earth Science’ at the ACM/IEEE SC21 conference.

Undergraduate Research Fellow

National Science Foundation

• Developed an image classification service deployed on AWS to enable rapid sea ice processing for climate and cryosphere research.
• Engineered UNet, FCN, and DeepLabV3 semantic segmentation models trained on high spatial resolution sea ice imagery.
• Implemented a novel deep learning semantic segmentation model pipeline to improve sea ice classification accuracy by 36%.
• {"Contributing author for ‘ArcCI"=>"a sea ice high resolution aerial image management and processing platform’ in Recent Advancements in Geoinformatics and Data Science (GSA Books)."}

Research Intern

United Nations

• Engineered semantic networks using a NoSQL Neo4j graph database to correlate data from the GTD, sanction lists, Panama Papers, Twitter, and crypto wallets to learn hidden patterns and analyze connections between entities in terrorist networks.
• Applied univariate time series forecasting on GTD dataset to predict if a terrorist attack might occur within a given timeframe.
• Modeled AMLSim synthetic banking transaction data on graph convolutional networks and XGBoost for the UN Office of Counter-Terrorism’s goFintel platform to detect and counter illegal money laundering and terrorism financing activities.

Research Intern

Harvard University Center for Geographical Analysis

• Documented 100+ key community resources & underlying infrastructure important for the mitigation of COVID-19 impacts.
• Produced a literature review of 1,298 papers on the state & utility of geospatial analysis tools in the COVID-19 research community.
• Conducted statistical NLP data analysis with NLTK, matplotlib, NetworkX, & pandas to identify the utility of geospatial analysis tools.
• Contributing author for ‘Quantitative geographical approaches in COVID-19 research - A review on first- and second-order impacts’ in Geospatial Stories of the Global COVID-19 Pandemic (Springer Nature).

Video Models for Efficient Disease Detection using Echocardiograms

• Tested various state-of-the-art DNN video model architectures & pre-training approaches to improve the training efficiency and overall accuracy of video-based disease detection models. Developed with the UCSF Tison Lab.

Modelling Amyloid Beta Plaque Formation in Alzheimer’s Disease

• Explainable CNN & VAE models learning the relationship between plaque morphology & molecular variation & determining the presence of amyloid plaques (protein aggregates). Developed as part of The Alan Turing Institute’s Data Study Group on Alzheimer’s Research alongside the UK DRI & DEMON.

Citation.ai

• Web application built using Python, React, & the OpenAI GPT-3 API to generate accurate & efficient citations following APA guidelines for researchers, students, & research librarians. Developed with support from the Harvard Undergraduate Machine Learning Research Lab & Harvard Innovation Labs.

Telepath Application

• Mobile application using deep autoencoders for video compression coupled with hardware acceleration to allow for people with unstable internet connections to remain connected with remote healthcare professionals. Developed for the Cal Hacks hack:now hackathon, winner of the IBM & Clinton Global University award.

Youth Delegate, inaugural U.S. Youth Consultation for UN Climate Strategy

• United Nations Foundation, United Nations Environmental Programme

The Leadership Award

• Cal Alumni Association

Unite Ideas Financial Crime Data Challenge Winner

• United Nations

2021 FORM+FUND Fellow

• Startup[at]BerkeleyLaw

2021 UN Millenium Fellow

• United Nations Academic Impact and MCN

ThinkChicago Chicago Ideas 2020 Recipient

• World Business Chicago

2020 Columbia-MIT-Harvard Summer Accelerator Fellow

• Almaworks Accelerator

ThinkChicago-from-Home 2020 Recipient

• World Business Chicago

IBM CGI U at hack:now

• Cal Hacks