Yifeng Shi

Principal Architect & AI Research Lead on Computer Vision

Autonomous Driving Foundation Models Spatial World Models Agentic Visual Model Self-Training

Email: shiyifeng@tju.edu.cn | Google Scholar: Profile

Education B.Eng. & M.Eng. in Information Engineering, Tianjin University

Key service CSIG Council Member NeurIPS 2026 Area Chair ACMMM 2026 PC ECCV 2024 CoopInt Workshop PC

Current KE Holdings (Spatial Intelligence / World Models) · Formerly Baidu (Autonomous Driving & Visual Foundation Systems)

About

I focus on building production-grade visual intelligence systems from research to large-scale deployment. My recent work centers on spatial intelligence and real-world 3D generation/world models, while also extending to cooperative autonomous driving and agentic visual model self-training.

Experience

Career Highlights

KE Holdings (2024.11 - Present): Principal Architect / Spatial Intelligence Lead. Leading the transition from conventional 3D reconstruction to world-model-driven spatial intelligence, with a focus on semantic understanding, physical reasoning, and generative 3D assets as core infrastructure for next-generation immersive living experiences.
Baidu ACG (2023.08 - 2024.11): Visual AI Chief Engineer / Leader. Replaced workshop-style custom delivery with an agentic closed-loop data-model pipeline, evolving visual AI development into a self-improving industrial system with stronger long-tail adaptation and cross-domain transfer efficiency.
Baidu IDG (2017.03 - 2023.08): V2X Autonomous Driving Perception Lead. Advanced the paradigm from single-agent perception to multi-agent cooperative intelligence and drove a unified vision foundation architecture for vehicle-road systems, validating foundation-model methodology in complex physical-world autonomy.

All Research Works (Chronological)

All papers are rendered from Scholar export in reverse chronological order, with unified Problem / Method / Impact cards.

Patents

Standards & Professional Activities

Key roles

CSIG (China Society of Image and Graphics): Council Member
NeurIPS 2026: Area Chair
ACMMM 2026: Program Committee
ECCV 2024 Cooperative Intelligence Workshop: Program Committee

Community & standards

Reviewer (selected journals): TPAMI, IJCV, TCSVT
Reviewer (selected conferences): CVPR, ICCV, ECCV, NeurIPS, AAAI, ACM MM
Vehicle-road data exchange standards & equipment testing practices
Contributor to vehicle-road cooperative ITS white papers
CVPR 2023 Foundation Model Workshop Challenge: Organizer
Greater Bay Area 3D Detection Challenge: Problem Setter