Cedrick Argueta

I work on machine learning at The Aerospace Corporation. Previously, I was a graduate student at Princeton University with Jaime Fernández Fisac. Prior to graduate school, I was an undergraduate at Stanford University.

During summers in college, I worked at The Aerospace Corporation and NASA JPL. I am an amateur astronomer and loved contributing to work that coincided with my hobbies.

Outside of research, I enjoy the outdoors, fashion, sports, and photography. I'm an avid runner and cyclist, used to collect sneakers, root for the Lakers, and like taking pictures of flowers.

Email  /  GPG  /  CV  /  GitHub  /  LinkedIn  /  Twitter

I'm primarily interested in reinforcement learning, especially in the context of robotics, safety, alignment, and language.

🪑

Large language models already reproduce falsehoods from human language, but are they actually convincing? We design a benchmark for future language models to test against, using tools from psychology and information theory to quantify exactly *how* convincing a language model is. This benchmark is part of the Google BIG-bench project.

A short thesis on what I've learned while completing honors at Stanford. I did my undergraduate honors thesis in the Stanford Intelligent Systems Lab with Prof. Mykel Kochenderfer and Louis Dressel, who is now at Wing.

A simulation package for the bearing-only localization problem, used in my reinforcement learning research.

Some photos that I've taken.

Visualizations of various strange attractors, created using Python. Check out the source code!

A contextual bandit problem of prescribing warfarin to patients, solved using LinUCB and online supervised learning. Check out our report! Work done for Stanford CS 234.

An agent that uses twin delayed DDPG for continuous control in bearing-only localization. Check out my report! Work done for Stanford CS 229.

An extension of CycleGAN and pix2pix, where we use generative networks to turn drawings of trees into photorealistic versions of those drawings. Check out our report! Work done for Stanford CS 231n.

A project using HFT techniques and statistical arbitrage on stocks in the NASDAQ-100 index. Work done for Stanford MS&E 448.

A simple agent that plays FlappyBird, created with Keras-rl and OpenAI Gym. Check out our report! Work done for Stanford CS 221.

A portable, automated testbed for safe-to-mate testing on CubeSat avionics boards. Work done at NASA JPL.

Aug. 2020 - May 2022, Department of Computer Science, Princeton University

M.A. in Computer Science

Sept. 2016 - Jun. 2020, Department of Computer Science, Stanford University

B.S. w/ Honors in Computer Science
Study abroad in Madrid, Spain.

Sept. 2012 - Jun. 2016, Science, Technology, and Mathematics Magnet, Abraham Lincoln High School

High School Diploma