Ethan Lai

EE + CS @ Stanford — building at the intersection of Physical AI, world models, and low-latency systems.

NOW

▼ press start

About

I'm an undergrad at Stanford University studying Electrical Engineering and Computer Science. I'm interested in Physical AI, world models, and low-latency, hardware-accelerated systems.

I like building things that bridge the gap between software and the physical world — whether that's quantized models for Physical AI or autonomous robots in space.

Most of my time goes into robots. Right now that means Deploy, a YC startup building production-grade robot deployments, and the Tambe Lab, where I'm getting vision-language-action models small enough to run on the computer that sits on the robot. It's PyTorch in the morning and C in the afternoon. Last summer I was a forward-deployed engineer at Wedge, another YC startup, which mostly meant getting ML pipelines to run on old hospital servers without breaking anything HIPAA cares about. The thread through all of it is latency: making models small and fast enough for the hardware that is actually there.

If any of that sounds interesting, email me at ethanlai@stanford.edu.

Experience

  • Robot Learning Intern @ DeployYC F26

    Sep 2026 – Present

    Early-stage startup building production-grade robot deployments for enterprises.

  • Physical AI Researcher @ Tambe Group, Stanford

    Feb 2026 – Present

    Custom firmware and memory architectures for VLA models

  • Forward Deployed Engineer @ WedgeYC S25

    Aug – Nov 2025

    Early-stage startup building ML infrastructure for healthcare systems.

Projects

4 wins
  • Rope manipulation rollouts from the latent world model: initial, planned, and final U-shape states

    lv 01

    Latent World Modeling

    CS 231n Final Project

    A compact JEPA-based latent world model for deformable-object manipulation from RGB observations.

    Paper

    • Deep Learning
    • World Models
    • Robotics
  • Inside the theremin: breadboards, a Mango Pi, two speakers, and ultrasonic sensors in a white box

    lv 02

    Bare-Metal Embedded Theremin

    CS 107e Final Project

    A bare-metal musical theremin on Mango Pi with real-time sensor input and I2S audio output.

    PhotosVideo

    • Embedded
    • C
    • Audio
  • lv 03

    CANOPY

    Army xTech 3rd National Security Hackathon

    2nd Place + $20,000

    An autonomous drone-satellite space communication system for national security.

    CodeYouTube Demo

    • Space
    • Drones
    • Comms
  • lv 04

    RECON

    YC RoboHacks 2026

    2nd Place

    An autonomous bomb defusal robot for rescue scenarios.

    CodeDemo

    • Robotics
    • VLM
    • ROS2
  • lv 05

    Detour

    TreeHacks 2026

    NVIDIA Edge AI Winner

    An autonomous collision avoidance system for space debris.

    CodeDevpost

    • AI
    • Space
    • Python
  • lv 06

    Anatomist

    CalHacks 12.0

    AI Design Track Winner

    An AR-based 3D anatomy app for real-time rendering.

    CodeDevpost

    • AR
    • 3D
    • Education

Stack

Technical

  • Isaac Lab
  • MuJoCo
  • ROS2
  • Raspberry Pi
  • RISC-V
  • CUDA
  • LangChain
  • Fusion 360
  • Onshape
  • MATLAB

Programming

  • C
  • Python
  • C++
  • PyTorch
  • Verilog
  • RTL
  • NumPy
  • SQL
  • React
  • Next.js