Robotics Researcher @ University of California, Berkeley Jun. 2026 - Present
Building C++/ROS 2 trajectory-prediction software for DARPA's TIAMAT program to anticipate rival autonomous vehicle motion.
Aug. 2025 – May 2026
Concentration Area: Signal Processing
Activities: Product Design Affinity Group Leader; DARPA TIAMAT Researcher
Relevant Coursework: Information Theory and Coding; Learning-Enabled Multi-Agent Systems; User Interface Design and Development; Advanced Natural Language Processing; Building User-Centered Programming Tools; Project Management.
Aug. 2022 - Jun. 2025
GPA: 3.97
Activities: Chancellor's Intern; Writing Teaching Assistant; College Scholars Program - Undergraduate Honors; Artificial Intelligence Club, Robotics Club, and Creative Writing Club; UCSC Website Manager
Relevant Coursework: Data Structures and Algorithms; Advanced Python; Assembly; C Programming; Computer Architecture; Statistical Probability; Discrete Math; Linear Algebra; Calculus I-III.
Aug 2018 - Jun 2022
Activities: Yoga Club, Artificial Intelligence Organization, Literary Magazine
Building C++/ROS 2 trajectory-prediction software for DARPA's TIAMAT program to anticipate rival autonomous vehicle motion.
Built a carbon-aware scheduling platform with Siemens Building X and a Python/FastAPI MILP optimizer, reducing modeled emissions by 35%. Designed a React dashboard for constraint setup, schedule overrides, and audit logs.
Built a Python/REST automation framework in VMware that cut virtual network-monitoring deployment time by 95%. Authored enterprise integration documentation for customer deployments.
Developed Monte Carlo simulations for Guardian Sensor 4 CGM with 300+ blood-glucose calibration variations. Used MATLAB and YAML configuration to improve insulin-delivery algorithm accuracy and built Simulink test coverage.
Embedded historical support issues in a Pinecone vector database and built a React/Twilio chatbot to surface issue frequency, bug classification, and debugging guidance.
AI-powered tool that tailors resumes and cover letters to specific job openings using Claude LLM API.
demo →MPC-to-PPO pipeline for autonomous highway merging with heterogeneous driver types, achieving 96% merge success in simulation.
code →Multimodal benchmark of 450 QA pairs across NLP lecture slides, with a ColPali retrieval-augmented VLM pipeline.
code →Domain-specific language for structured prior-art search over patents, with a FastAPI backend and React frontend.
code →AI-powered bug tracking using vector embeddings and semantic similarity to identify related GitHub issues.
code →Voice-first web app connecting senior citizens with volunteer companions using AI-powered onboarding and smart matching.
code →