Bio

Experience

Software Engineering Intern @ Warp – Mar. 2023 to Present

Undergraduate Researcher @ Berkeley RISE Lab – Sep. 2022 to Mar. 2023

Developed system to efficiently deploy and shadow-test large ML models on resource constrained edge devices under PhD Student Shishir Patil and Prof. Joseph Gonzalez.
Added functionality to convert neural network models from Tensorflow, PyTorch, and SKLearn into C code by compiling from ONNX intermediate representation.

Software Engineering Intern @ Actively – Sep. 2022 to Nov. 2022

Designed and developed metric learning framework that leverages active learning principles (uncertainty and diversity sampling) to interactively finetune classification models using human feedback.
Created an Nginx reverse proxy to host custom Streamlit applications and added OAuth 2.0 user authentication.
Added regression support and inference optimizations to Bayesian Network based core predictive model.
Aided in engineering tasks and completed logistical tasks through Laika for achieving SOC-II compliance.

Undergraduate Researcher @ Berkeley RISE Lab – Mar. 2022 to Sep. 2022

Worked on developing a novel concurrency control protocol for long-lasting transactions in large-scale graph databases under PhD Student Audrey Cheng and Prof. Ion Stoica.
Created Python benchmarking tool for Neo4j to assess performance in 50M+ node graph creations and deletions.
Prototyped protocol ideas based on distributed sagas, altruistic locking, and multi-version concurrency control.

Software Engineering Intern @ PayPal – May 2022 to Aug. 2022

Developed an early warning notification system to alert merchants likely to incur fines under Fraud/Dispute Monitoring programs based on recent transaction activity.
Created a new Spring Boot microservice to execute asynchronous daily/weekly batch jobs.
Developed internal library that provides a flexible API for connecting to BigQuery data warehouse, quickly creating parameterized query strings, and running/monitoring synchronous and asynchronous query jobs.

Undergraduate Researcher @ CAHLR Lab – May 2021 to May 2022

Worked in the CAHLR Lab under Prof. Zachary Pardos. Developed a novel convolutional neural network architecture, ideated neural embedding techniques for courses, designed novel evaluation metric for prerequisite prediction performance measurement, and curated custom training dataset to devise a novel deep learning approach to course prerequisite prediction.

Machine Learning Fellow @ Launchpad.AI – May 2021 to Aug. 2021

Worked as a fellow in the 19th Cohort of Launchpad.AI’s Fellowship.AI program on two projects over the course of 4 months:
Masking Fetal Gender in Ultrasound Images:
- Developed a cutting-edge real-time object detection system to mask fetal gender in ultrasound videos of pregnant women.
- Optimized YOLOv5 for ultrasound object detection and implemented structural similarity-based technique to minimize false negatives during video inference.
- Co-authored research paper detailing our effort; Pre-print released on arXiv.
Fitness Chatbot Assistant:
- Developed a MVP for a chatbot assistant for athletes that is specialized to answer fitness, diet, and sports related queries.
- Leveraged RASA Framework to build and finetune chatbot and added integrations with serpAPI Google Search and BlenderBot to handle out-of-scope queries.
Worked in a fast-paced result-oriented environment, engaging in weekly sprints and presenting to the CEO on a weekly basis.

Machine Learning Intern @ CAHLR Lab – Jan 2021 to May 2021

Worked under Prof. Zachary Pardos in the CAHLR Lab as an ML Developer on developing the course prerequisite prediction functionality of the AskOski Research Project (https://askoski.berkeley.edu) and improved the feature’s Recall@10 metric by 5%.
Developed novel multi-input autoencoder based neural network architecture that combines a BERT encoder and Course2Vec vector encoder that is capable of predicting multiple prerequisites for a given course.
Utilized 900,000+ row student enrollment history dataset and ~10,000 row course catalog description dataset in training.
Iteratively improved solution over the course of the project, experimenting with ensembles of classical ML models, other neural network architectures, and other NLP vector representation techniques like Doc2Vec along the way.

Data Consultant @ TripAdvisor – Feb 2021 to May 2021

Developed NLP based solutions for TripAdvisor working through Big Data at Berkeley’s Projects Committee.
Implemented BERT based neural network for multi-class multi-label classification of customer service chat interactions that achieved an accuracy of 92%.
Leveraged autoencoder based approach for anomaly detection in order to create an issue novelty flagging system.
Both classification and anomaly detection models deployed and actively used in production.
Utilized unsupervised topic modelling techniques on raw chat corpus to extract and present valuable business insights regarding customer service workload optimization to Tripadvisor customer service team leadership.
Conducted extensive data cleaning, text data preprocessing, feature generation, and text embedding generation.

Software Engineering Intern @ PointR Data Inc. – May 2019 to Aug 2019

Developed multiple-object tracking algorithm to track and assign unique IDs to customers in real-time from raw CCTV camera footage.
Leveraged Tiny YOLOv3 Architecture for object detection task; conducted transfer learning in order to optimize human detection accuracy.
Implemented object tracking methods based on bounding box velocity and frame-to-frame bounding box IOU.

Computer Science

EECS

Mathematics

Sciences

Breadth

SOCIOL 1 - Introduction to Sociology (Sp23)
HISTORY 6B - Introduction to Chinese History (Sp23)
EECS 194 - Social Justice in EECS (Sp23)
EPS 109 - Computer Simulations with Jupyter Notebooks (Fa22)
HISTORY 137AC - Immigrants and Immigration as US History (Sp22)
COLWRIT R4B - Reading and Composition Part B: The Machine and its Discontents (Sp22)
SOCIOL 167 - Virtual Communities and Social Media (Fa21)
ECON 1 - Introduction to Economics (Su20)
COLWRIT R4A - Reading and Composition Part A: Texts of the Apocalypse (Su20)