Aaron Trinh

Hi! I'm a third year Computer Science undergraduate student at Georgia Tech, where I've been fortunate to work with Prof. Sehoon Ha and Prof. Bo Dai.

My research interest is in enabling learning of intelligent agents that can reason about the world. To this end, I am interested in foundation models for decision making, world models, and reinforcement learning.

atrinh31 [at] gatech [dot] edu  /  Google Scholar  /  Github  /  LinkedIn  /  Twitter

profile photo

Research

(* indicates equal contribution)
Image Reference Grounded Skill Discovery
Seungeun Rho, Aaron Trinh, Danfei Xu, Sehoon Ha
International Conference on Learning Representations (ICLR), 2026
arXiv

Image Towards Better Instruction Following Retrieval Models
Yuchen Zhuang*, Aaron Trinh*, Rushi Qiang*, Haotian Sun, Chao Zhang, Hanjun Dai, Bo Dai
arXiv, 2025
arXiv / code / data

Projects
Image Efficient Skill-based Reinforcement Learning
Course Project CS 8803 - Deep Reinforcement Learning
code

A model-based RL framework that extracts reusable skills from rewardless offline data and reuses the data by relabeling with an optimistic reward estimator for efficient exploration through the learned dynamics model

Template from Jon Barron.