About Me

I am a research scientist at Luma AI, currently focused on post training large multi-modal models. I was a core contributor to Luma's Ray3 model, for which I lead supervised fine tuning (SFT). In general, I am interested in the intersection of data and model design for large AI systems. I received my PhD in Computer Science from Stanford University in 2025, where I was advised by Leonidas Guibas. During my PhD, I focused on training AI models for better understanding and reconstructing the 3D and 4D world we live in. I received my Bachelor's degree in computer science from MIT in 2020. Over the years, I have also had the opportunity to intern with and/or collaborate with a variety of companies, including Google, Apple, Nvidia, and Toyota.

Research

Luma Ray 3

The Team at Luma AI. 2025.

Dynamic Gaussian Marbles for Novel View Synthesis of Casual Monocular Videos

Colton Stearns, Adam W. Harley, Mikaela Uy, Florian Dubost, Federico Tombari, Gordon Wetzstein, Leonidas J. Guibas. SIGGRAPH Asia 2024.

• Project Website
• Paper
• Code

View-Consistent Hierarchical 3D Segmentation Using Ultrametric Feature Fields

Haodi He, Colton Stearns, Adam W. Harley, Leonidas J. Guibas. ECCV 2024.

• Paper
• Code

CurveCloudNet: Processing Point Clouds with 1D Structure

C. Stearns, D. Rempe, Jiateng Liu, Alex Fu, Sebastien Mascha, Jeong Joon Park, Despoina Paschalidou, Leonidas J. Guibas. CVPR 2024.

• Paper
• Code

SpOT: Spatiotemporal Modeling for 3D Object Tracking

C. Stearns, D. Rempe, J. Lie, R. Ambrus, Y. Yang, S. Zakharov, V. Guizilini, and L. Guibas. ECCV 2022 Oral Presentation.

• Project Website
• Paper
• Code