Parth Diwane

I research embodied AI, multi-agent collaboration, AI agents, and reinforcement learning. These days I'm a SWE intern @ Cisco and researcher at the NLP Group. In my free time I like hiking, watching anime, listening to music, and working out.

Interests

NLP · Embodied AI · RL · LLM Risk Analysis

School

UC Santa Barbara

Program

Computer Engineering

GPA

4.0

Deans list

X2

experience

[research & industry]

Oct 2025 - March 2026

Amazon

AI Researcher

Worked on embodied AI benchmarks involing multi-agent collaboration and theory of mind

Oct 2025 - Present

NLP Researcher

UCSB NLP Group - Advised by Xin Eric Wang

Jun 2024 - Jan 2025

Georgia Tech

Computer Networking Researcher

Researched BGP protocol and hijacking attacks on BGP.

selected builds

[projects]

01

EmToM

benchmark for multi-agent collaboration in an embodied (Partnr) setting.

02

card-counter

counting cards using CV, Meta glasses, and the MiDaS model.

03

match-point

predicting outcomes of tennis matches using random forests.

04

maze-agent

reinforcement learning project for training an agent to navigate a maze environment.

publications

[research]

EmToM: Embodied Agent Theory of Mind Evaluation Benchmark

In Progress

Gurusha Juneja, Dylan Lu, Saaket Agashe, Parth Diwane, Xin Eric Wang

writing

coming soon

contact

You can reach me at email and LinkedIn, or stalk my projects at GitHub :)