hey! my name is

Aadi Shah

Aadi Shah

I'm an undergrad at Stanford studying computer science and philosophy. I'm broadly interested in memory and continual learning for generalizable robot learning policies at the Stanford AI Lab, where I'm part of the IRIS group advised by Chelsea Finn. I previously spent a summer at Google working on Gemini Enterprise for CX.

telemetry
this arm is a live RL policy, running in your browser!→

research

  • Combining Symbolic and Perceptual Memory for Long-Horizon Robot Manipulation
    Aaditya Shah
    CS231N: Deep Learning for Computer Vision
    Nominated for Best Paper Award
  • RECAP-Ψ: Advantage-Conditioned Fine-Tuning for Open-Source Humanoid VLAs
    Jonathan Lu*, Aaditya Shah*, Karthik Pythireddi*
    CS224R: Deep Reinforcement Learning
  • SLLU pipeline diagram: domain documents are embedded, retrieved, and combined with a refinement dialogue to prompt an LLM that generates sUAS test scenarios.
    Agentic LLM-Driven Simulation Testing and Automated Analysis for Small Uncrewed Aerial Systems
    Jithin Garapati, Aaditya Shah, Bohan Zhang, Michael Vierhauser, Ankit Agrawal
    Preprint (SSRN)

projects

  • Crate-stacking robot arm
    A pixel robot arm driven by a PPO policy I trained, running live in the browser.
  • Predator–Prey Simulator
    A C++ simulation where predator and prey agents are driven by neural networks evolved over generations, so hunting and evasion emerge from selection alone.
  • krates
    Your AI devops engineer before you can afford to hire one — automating Docker and Kubernetes for early-stage teams.
    CalHacks 12.0
    1st place, YC Track

Last updated: October 2026

aadi2 [at] stanford [dot] edu