About

I’m a fourth-year Electrical Engineering and Computer Science student at the University of California, Berkeley, originally from Palo Alto, CA. My experiences span professional software engineering, machine learning, and academic research, centered on building scalable systems and translating technical complexity into user-driven solutions. I am especially interested in the intersection of artificial intelligence and product development, particularly in building intelligent systems that drive real-world problem-solving and enhance user experiences. As I complete my final semesters here at UC Berkeley, I’m eager to continue working on high-impact projects and expanding my technical knowledge. Thanks for stopping by, please feel free to reach out or connect with me at one of the links to the left!

Education

University of California, Berkeley • 2022-26

Electrical Engineering and Computer Science (BS)

Data Science (Minor)

Experience

Software Development Engineer Intern

Built and deployed a fully automated test coverage analysis system for the Alexa+ testing platform (ATO), improving observability across 2,000+ Alexa+ APIs and LLM's by 100% using serverless AWS architecture. Accelerated LLM evaluation workflows through an event-driven, real-time trace analysis pipeline (DynamoDB, SNS, SQS, Lambda, S3), providing immediate API-level coverage metrics as well as daily test data aggregation for thousands of test runs per day. Implemented ad hoc querying and data exploration infrastructure with Amazon Athena and QuickSight, enabling Alexa teams to query and visualize millions of test records across months of Alexa+ testing to track API coverage trends over time and prioritize future testing development. Led the end-to-end delivery of this system—including PoC, technical design, and full-scale production deployment, ensuring operational scalability and seamless integration with Alexa-wide ML testing infrastructure and data pipelines.

Research Assistant

Built a propaganda and misinformation detection pipeline by training initial classifiers on a custom-labeled dataset of 600+ Ukraine-related articles using fine-tuned, transformer-based models such as RoBERTa and DistilBERT. Improved narrative detection accuracy by 30% through iterative prompt engineering using GPT-4, Claude, and Mistral models to identify narrative types and variant patterns from multilingual content. Additionally, automated model evaluation and batch inference across thousands of online articles by integrating tools for JSON-based output formatting, confidence scoring, and reasoning traceability.

Software Engineer Intern

Built a prototype recommendation engine to personalize local event suggestions, resulting in a 15% increase in platform engagement based on CTR metrics. Used scikit-learn to implement a content-based filtering model leveraging event categories, tags, and location and cleaned and processed over 5,000 community event records from city open data APIs, standardizing metadata such as dates, venues, and categories to enable accurate filtering and effective model training.

Research Assistant

Worked on a joint research product with UCSF building nanotechnology-based portable blood sampling devices. Contributed to the development of the team’s supervised machine learning model (PyTorch) trained on historical blood sample data to predict potential health conditions and recommend future diagnostic tests from the results of their devices.

Projects

SIXT33N: Built a voice-controlled model car by integrating machine learning techniques (PCA, SVD) with C++ programming to process four voice commands; implemented system on a microcontroller, utilizing analog circuits and custom PCBs for hardware control, engineered op-amp and BJT circuits for closed-loop control logic, and audio signal amplification and filtering circuits to process the voice commands.

ANN Network: Wrote RISC-V assembly code to execute a simplified artificial neural network (ANN) on the Venus RISC-V simulator; built and integrated fundamental operations such as vector dot products, matrix-matrix multiplications, argmax computations, and activation functions to load and execute a pre-trained network to classify handwritten digits from the MNIST benchmark set.

Drive: Designed and built a secure file sharing system (based on Google Drive) in Go with rootKey-based authentication, cascading access revocation for dynamic permission updates, integrity validation, and delegated access control using encryption keys and file location pointers; supported user creation/deletion and file saving, editing, loading, and sharing/revocation.

CPU: Built a fully functional, 32-bit CPU with a pipelined RISC-V datapath in Logisim capable of running machine code converted from RISC-V assembly code; supports efficient instruction processing and adheres to the RISC-V ISA specifications.

NGram: Replicated Google’s Ngram Viewer in Java using HashMaps, Collections, Iterators, and the Princeton Algorithms Library for reading, parsing, and converting text datasets to track history of word usage, enabling query handling and scalability for linguistic data processing.

WordNet: Built a visual and numerical model of hyponym and hypernym correlations and tracked word tree lineage in Java, using Directed Acyclic Graphs, HashSets, and ArrayLists, as well as algorithms including depth-first search graph traversal and weighted quick union.

2D World Generation: Designed and implemented the game functionality, UI, and data storage of a 2D tile-based, user-interactive world exploration engine in Java that uses pseudo-random algorithms to generate custom worlds and gameplay.

Scheme: Implemented the core features for a lisp interpreter in Python using a recursive descent parser, utilizing lexical and syntactic analysis as well as input parsing.

Arcade Games: Built fully rendered replications of the classic puzzle games 2048 and Tetris in Java that incorporate user data storage for scores, responsive user controls through keyboard inputs (WASD, UDLR), and various Java data structures and algorithms for game logic.

Website: Built using HTML and CSS, hosted on GitHub Pages. View the code here.

Skills

Languages: Python, Java, C/C++, Go, SQL, RISC-V, Swift, HTML, JavaScript, CSS

Developer Tools: Git, Google Cloud, VS Code, IntelliJ, Xcode, Figma, SwiftUI

Libraries: NumPy, Matplotlib, pandas, React, React Native