2026
Boogu-Image-0.1: Boosting Open Agentic Multimodal Generation via Understanding under a Minimal Budget
Content-safety safeguards for responsible text-to-image generation and instruction-based editing.
AI Research Engineer
Visual generation, world models & interaction, and AI infrastructure.
About
I am Chong Hou Choi, an AI researcher currently working on world models and real-time visual interaction. Previously, I was an AI Algorithm Engineer at Huawei's Hong Kong Research Center, where I had the opportunity to work with Dr. Rui Liu ↗. Before that, I was a Machine Learning Engineer Intern at Shanghai Artificial Intelligence Laboratory ↗ under the supervision of Dr. Yi Wang ↗.
My broader interests include visual generation, interactive AI systems, and AI infrastructure. I am particularly interested in exploring how visual intelligence can support education and learning.
Selected publications
032026
Content-safety safeguards for responsible text-to-image generation and instruction-based editing.
2026
A data construction and quality-control system for approximately 100K aligned source-target video pairs.
2026
A shared data and inference pipeline spanning image and video generation and editing.
Experience
022024 — 2026
Worked across video understanding, large-scale data construction, image restoration, multimodal generation, and model evaluation. Built a DINOv3-based shot boundary pipeline that automated the cleansing of more than five million video clips.
2022
Optimized temporal action localization for INTERN 2.0 and improved performance on major benchmarks.
Education
022023 — 2024
Master of Engineering in Electrical and Computer Engineering
Waterloo, Canada
2019 — 2023
Bachelor of Engineering in Automation Engineering
Beijing, China
Places I wander, music I love, and people I'm lucky to have beside me.

