cv

SDE/ML Intern

Aurora Innovation

Optimizing dynamics model in simulation using DEC deep learning clustering approach where the input is a dataset of log snippets where sim2real gap is high.

Built an end-to-end pipeline in Python to ingest full on-road logs (roughly 3 hrs), run parallel simulation, and extract snippets with significant sim2real gap. This pipeline is already in production after just 1 week at my internship.

Currently architecting and training clustering model by pre-training a LSTM Autoencoder and feeding latent vector to Deep Embedded Clustering (DEC) model with the explicit focus on clustering.

• Deep Learning
• Simulation
• Autonomous Driving
• Probabilistic Robotics

Student Researcher (PRIOR Team)

Allen Institute for AI (Ai2)

Built a bimanual system of FR3 robotic arms. Currently researching on benchmarking different policies for fine manipulation on the real-world system. Developed keyboard-based and VR-based teleoperation software using MoveIt motion planning & inverse kinematics in Python.

Benchmarking paper is published and in review for CoRL 2025. Read the paper here.

• Robot Learning
• Foundation Models

Undergraduate ML Researcher

Personal Robotics Lab @ UW

Tested imitation learning control algorithm (CCIL) on real 6-DOF robotic arm & simulation (F1Tenth, Drone, Mujoco, Metaworld). CCIL outperformed human and top algorithmic baselines in both real and simulated environments by 75%. Performed robot-learning research for fine manipulation on 6-DOF robotic arm and miniature autonomous vehicles.

• Machine Learning
• Imitation Learning
• Robot Learning

Bachelors of Arts, Bachelors of Science

University of Washington, Seattle

Computer Science, Applied Math, Economics

• GPA: 3.89/4.00
• CSE 478 - Autonomous Robotics
• CSE 493G1 - Deep Learning
• MATH 407 - Linear Optimization
• CSE 332 - Data Structures and Parallelism
• CSE 333 - Systems Programming
• CSE 589 - Software Entrepreneurship (Grad-level)
• CSE 351 - Hardware/Software Interface
• CSE 311 - Foundations of Computing I
• CSE 312 - Foundations of Computing II
• AMATH 352 - Applied Linear Algebra and Numerical Analysis
• MATH 208 - Linear Algebra
• MATH 12X - Calculus I-III
• STAT 311 - Elements of Statistical Methods
• AMATH 301 - Intro to Scientific Computing
• ECON 20X - Macro/Micro Economics
• CHEM 1X2 - General Chemistry I-III

RoboEval: Where Robotic Manipulation Meets Structured and Scalable Evaluation

arXiv, in submission to CoRL 2025

RoboEval is a simulation benchmark and structured evaluation framework designed to uncover specific weaknesses in bimanual manipulation policies. Unlike traditional binary success metrics, RoboEval includes tiered, skill-based tasks that systematically test spatial, physical, and coordination skills, alongside fine-grained diagnostic metrics and over 3,000 human demonstrations for imitation learning. Experiments show that seemingly successful policies differ significantly in their execution quality, with behavioral metrics highlighting precise issues such as alignment and temporal consistency. RoboEval thus provides deeper, actionable insights into manipulation policy limitations beyond simple success rates.

7x Dean's List

University of Washington

The University of Washington recognizes students after each academic quarter and year for their high scholarship award and inclusion on the corresponding Dean’s List.

Bimanual Teleop Software

Inverse kinematics-based teleoperation software for an Meta Quest 2 VR controller for a bimanual FRANKA arm system.

• Python
• Robotics

Discover

All-in-one chest x-ray analysis, diagnosis, and catheter placement web application using AI. Developed an image classification and segmentation model (over 60M param) with TensorFlow, OpenCV, and PyTorch, trained on thousands of x-rays. Front-end built with Flask, Python, HTML/CSS. Won award in CodeDay Hackathon 2023.

• Python
• TensorFlow
• OpenCV
• Flask
• Image Segmentation
• Image Classification

Human Image Masking (Segmentation)

Learned how to create a Image Seg model with U-NET architecture and developed and implemented the model to segment and mask human presence in images. Also learned data augmentation using Albumentations for training a better model.

• Python
• PyTorch
• TensorFlow
• OpenCV
• Albumentations
• Image Segmentation

Mach 5 Robotics

Collaborated with peers to write teleop & auton code for 22-23 VEX high school competition.

• C++
• Robotics