cv
Short Bio
I have experience in Vision, RL, Generative Models, Sequence Models, Self-supervised Learning, GCNs, LLMs. My PhD thesis revolves around multiple aspects of LLM agents, including pretraining, evaluation, and prompting architectures.
Education
- 2021 - Current
PhD in Machine Learning
Carnegie Mellon University, Pittsburgh, PA, USA
- Advisor: Dr. Tom Mitchell, Dr. Yuanzhi Li
- 2017 - 2020
B.S. in Computer Science
Carnegie Mellon University, Pittsburgh, PA, USA
- B.S. in Computer Science (Major GPA 3.91/4.0)
- Minor in Machine Learning (Minor GPA 4.0/4.0)
- Department Honors and University Honors
- Alumni Award for Undergraduate Excellence
Research Experience
- Summer 2024
Quant Research Intern
Point72 Cubist
- LLM for trading strategies.
- Fall 2023
Consultant
Microsoft Research | Host: Dr. Yuanzhi Li and Dr. Sébastien Bubeck
- GPT and LLAMA. LLM agents for games
- Summer 2023
Research Intern
Microsoft Research | Host: Dr. Yuanzhi Li and Dr. Sébastien Bubeck
- Summer 2022
Research Intern
IBM Research | Host: Yuya Ong and Dr. Taiga Nakamura
- Graph-based recommender-system that operates unstructured long texts with the use of large language models. SOTA at competitive benchmarks.
- 2021
Research Assistant
Machine Learning Department (in collaboration with Apple A.I. Research) | With Dr. Ruslan Salakhutdinov, Dr. Shuangfei Zhai, and Dr. Joshua M. Susskind
- Offline RL and Energy Based Models formulation for sequence generation.
- Summer 2020
Research Intern
Apple A.I. Research | Host: Dr. Hanlin Goh and Dr. Joshua M. Susskind
- Uncertainty estimation to stabilize actor-critic based offline reinforcement learning.
- Summer 2020
Research Assistant
MultiComp Lab, CMU | With Dr. Ruslan Salakhutdinov and Dr. Louis-Philippe Morency
- An information-theoretical explanation for the effectiveness of the unsupervised or self-supervised learned representations. The explanation inspires new loss function designs.
- Policy based unsupervised domain transfer framework for visual navigation. Obtains promising results on few-shot sim2real indoor navigation.
- 2019 - 2020
Research Assistant
Machine Learning Department | With Dr. Zhiting Hu and Dr. Eric P. Xing
- Novel GAN formulation through variational inference and constrained reinforcement learning that leads to probability ratio clipping and discriminator re-weighting.
- 2018 - 2019
Research Assistant
MultiComp Lab, CMU | With Dr. Louis-Philippe Morency
- Unsupervised Facial Landmark detection and structure-from motion.
- 2018 - 2019
Research Assistant
Biorobotics Lab, CMU | With Dr. Guillaume Sartoretti and Dr. Howie Choset
- Asynchronous Reinforcement Learning for collaborative robotic manipulation tasks.
- Asynchronous Reinforcement Learning for collaborative construction.
Media Coverage
- National Post
AI gamers benefit from reading the instruction manual first
by Chris Knight | 2023-03-20
- Singularity Hub
An AI Learned to Play Atari 6,000 Times Faster by Reading the Instructions
by Edd Gent | 2023-03-10
- New Scientist
AI masters video game 6000 times faster by reading the instructions
by Matthew Sparkes | 2023-03-04
Honors and Awards
- 2019
- School of Computer Science Alumni Award for Undergraduate Excellence
- $2000 Award at IBM BlueHack
- 2018
- Grand Award at HackAuton 2018
- Builders’ Choice and Staff’s Choice at Build18 2018
- 2017
- Most Entrepreneurial Hack at HackCMU
- Silver Tier at Halite 2 Competition