Hai Pham

Spatial Intelligence or die trying.

System Prompt

"### ROLE\n"
"You are an intelligent PhD student in Computer Vision and Graphics, specialized in making your advisor happy."
"Your goal is to make great contributions to outstanding vision problems and have fun while doing it."
"Follow your advisor, learn from friends and make sure you understand everything you are doing. \n\n"

· GitHub · Google Scholar · LinkedIn

11/26 –
(incoming)

To be announced soon ";)"

09/25 –
09/26

MSc. in the Mathématiques, Vision, Apprentissage (MVA) program at École Normale Supérieure Paris–Saclay. Also researching how to improve spatial reasoning for VLMs with Shizhe Chen and Cordelia Schmid.

04/25 –
08/25

Research Intern at Qualcomm AI Research, working on efficient 3D generation with diffusion transformers — including SharpDepth (CVPR 2025), which sharpens metric depth boundaries. Featured in a Qualcomm blog post.

09/23 –
03/25

Research at VinAI Research, supervised by Rang Nguyen, on making 3D scene understanding more label-efficient and robust to real-world data. Collaborated with Binh-Son Hua, Phong Nguyen, and Khoi Nguyen. This led to semantic scene completion work at AAAI 2025 and a multi-view occupancy challenge entry.

09/20 –
06/24

B.Eng. in Computer Engineering at Ho Chi Minh City University of Technology (Bach Khoa) – Vietnam National University.

bio

I am interested in the wonderful world of Computer Vision and Graphics. My goal is to develop simple systems that allow people to capture and simulate their environment to make art or (going with the trend here) train robots. Over the last few years, I've been working on multiple aspects of Vision including scene understanding and, more recently, reasoning.

publications

SharpDepth: Sharpening Metric Depth Predictions Using Diffusion Distillation
CVPR 2025
Duc-Hai Pham*, Tung Do*, Phong Nguyen, Binh-Son Hua, Khoi Nguyen, Rang Nguyen

paper project code

Semi-supervised 3D Semantic Scene Completion with 2D Vision Foundation Model Guidance
AAAI 2025
Duc-Hai Pham, Tuan Ho, Duc Dung Nguyen, Anh Pham, Phong Nguyen, Khoi Nguyen, Rang Nguyen

paper project

Also on Google Scholar for the canonical, auto-updated list.

pet projects

Smaller experiments on the side.

SwiftPolicy — score distillation on diffusion policies for visuomotor control, built for the MVA robotics course with Ianis Hammani.

3D Semantic Occupancy from Surrounding View Images — a CVPR 2023 3D Occupancy Prediction Challenge entry with Tuan Ngo, Tuan Ho, Khoi Nguyen, and Rang Nguyen. We predicted voxel semantics from multi-view images and ranked 16 / ~200 teams — at the time, the highest-ranked team from Southeast Asia on the leaderboard. See the challenge site.

Layout homage: karpathy.ai. Earlier template DNA from the academic web ecosystem; mistakes are mine.