Research

Research Intern at MIT DeCoDe Lab, Summer 2026

Developing multi-fidelity field prediction methods for PDE simulations and climate data using neural operators, convolutional networks, and ensemble methods. Built an autoresearch system to autonomously run experiments, analyze results, and iterate on model configurations. Mechanistically interpreted why the best-performing architecture outperformed alternatives. Second author paper submitted to ICLR 2027 main conference. SURF Report coming soon.

Research Intern at Caltech AMBER Lab, Summer 2025

This project focuses on integrating the Unitree D1 robotic arm into the Caltech AMBER Lab’s robotics research interface, Obelisk, with the goals of enabling loco-manipulation experiments on a quadruped-armed system. After extensive debugging of the Unitree D1 SDK and Obelisk, I developed a tele-operation system consisting of a controller, state estimator, and hardware interface. The controller performs inverse kinematics on the gripper, and I implemented a joystick-based interface for intuitive user control. Despite these efforts, the D1 arm failed to reliably follow commands, even after extensive troubleshooting and communication with Unitree Customer Service, ultimately leading to the conclusion that the hardware is defective. The project has pivoted toward preparation for future research in model predictive control and reinforcement learning.

Report

Research Intern at Caltech Daraio Group, Summer 2024.

I worked with Jabri Garcia-Jimenez ('26) to characterize the mechanical behavior of novel locking chains by conducting 140+ successful tensile load experiments using the Instron ElectroPuls 3000, achieving precise data on structural transitions. Post-doc Wenjie Zhou and PhD student Sujeeka Nadarajah were my mentors. Jabri and I split up the presentations/papers such that I would present the first half of our work and he would present the latter half.

SURF 2024 Group Presentation (No Videos) SURF 2024 Individual Presentation SURF 2024 Paper

Research Intern at USC Information Sciences Institute Laboratory for Quantum-Limited Information, June 2022-November 2022.

I designed and conducted experiments to estimate laser propagation direction using scattered light. Investigated classical Fisher information and developed deep neural networks that predicted laser angles with ±2.7º accuracy, offering a promising single-sensor alternative to traditional methods.

Poster Paper Specific Contributions

Researcher at MIT Beaver Works Summer Institute piPACT Independent Research Project Program, 2020.

Researched COVID-19 contact tracing feasibility using humidity, temperature, and Bluetooth signal strengths between two Raspberry Pis. Trained a logistic regression model with 57% test accuracy, highlighting sensor limitations for reliable proximity detection in summer conditions.

Paper Code