Hello! I currently work as a research engineer at Simile. Previously, I was an undergrad and master's student at Stanford studying computer science and math and did research with Michael Bernstein and Ada Aka.
My work bridges NLP and HCI, focusing on building computational models of human behavior from massive, open‑ended data sources. I've also worked on designing social‑media platforms to better empower end‑user autonomy. My goal is to create tools that transform our ability to reason deeply about people's actions, motivations, and decision‑making processes.
If you are interested in chatting about models of human behavior and their application to simulation, please reach out!
Work Experience
Simile
Member of Technical Staff
November 2025 - Present
Two Sigma Investments
Software Engineer Intern
June 2025 - August 2025
Stanford Human Computer Interaction Group
Research Assistant
June 2024 - Present
Stanford Graduate School of Business
Research Assistant
November 2023 - June 2025
Computational Media Lab @ University of Texas at Austin
Research Assistant
August 2021 - August 2023
Research Projects
Recent work is shown below. * denotes equal contribution.
Finetuning LLMs for Human Behavior Prediction in Social Science Experiments
Akaash Kolluri*, Shengguang Wu*, Joon Sung Park, Michael S. Bernstein
Alexandria: A Library of Pluralistic Values for Realtime Re-Ranking of Social Media Feeds
Akaash Kolluri*, Renn Su*, Farnaz Jahanbakhsh, Dora Zhao, Tiziano Piccardi, Michael S. Bernstein
Quantifying The Spread Of Racist Content On Fringe Social Media: A Case Study Of Parler
Akaash Kolluri*, Dhiraj Murthy*, Kami Vinton
Software Projects
Below are a few software projects I've worked on.
DebateV: An Evidence Database for Debaters
Made with two friends from debate camp, DebateV was a web interface for searching 1,000,000+ pieces of evidence for debaters to use to write their cases. While active from August 2020 to August 2023, it gained over 1,000,000 page visits.
Python Package: genagents
Worked as a developer on a Python package to enable researchers to easily build LLM‑based generative agents. The package accompanied the release of the paper Generative Agent Simulations of 1,000 People.