詹锟 | Kun Zhan
Head of Foundation Models & Autonomous Driving at Li Auto
Email: zk_1028@aliyun.com
WeChat: KevinZhan1990
Beijing, China
Building the vehicle-scale AI stack from perception to language to action
I lead Li Auto's foundation-model and autonomous-driving efforts, building the Mach VLA and Mach Mind model families and deploying them across production vehicles as reliable physical-world intelligence systems.
About Me
I'm Kun Zhan, Head of Foundation Models and Autonomous Driving at Li Auto. My work focuses on unifying machine intelligence and language intelligence into production-grade embodied systems: Mach VLA for three-dimensional perception, decision, and action, and Mach Mind-Pro / Mach Mind-Edge for cloud and on-device agentic reasoning.
Since joining Li Auto in 2021, I have been responsible for core autonomous-driving architecture evolution, from Highway NoA and City NoA to end-to-end, VLM-assisted, and VLA-based systems. Today my agenda spans model research, data engines, reinforcement learning, world models, chip-model co-design, on-vehicle inference, and OTA deployment at automotive scale.
Earlier, I earned a master's degree in Automation from Beihang University and led behavior-prediction work at Baidu Apollo, where I built large-scale motion-forecasting and planning systems for L4 autonomous-driving programs.
My mission is to build physical-world AGI, using autonomous driving as the starting point and expanding toward robots, intelligent spaces, and broader real-world embodied intelligence.
Key Highlights
Leadership, model releases, and production execution across the full embodied-intelligence stack.
Unified AI leadership
Lead Li Auto's foundation-model and autonomous-driving roadmap across Mach VLA, Mach Mind, 3D ViT, world models, reinforcement learning, model infrastructure, and on-vehicle deployment.
Mach model releases
Drove the release of the Mach VLA and Mach Mind-Pro / Mach Mind-Edge model families, connecting language intelligence with machine intelligence for vehicles, cabins, and future embodied products.
Full-stack production
Translate research systems into user-facing capability through model-chip-OS-domain-controller integration, including deployment paths around Li Auto's self-developed Mach M100 AI chip and Livis vehicles.
Recent Milestones
Public releases and talks that reflect the current direction of my work.
Livis Day keynote
Main speaker for Li Auto's software and embodied-intelligence launch, presenting the Mach model stack and the path from intelligent vehicles to embodied-intelligence products.
Watch event replayMach VLA
Announced the next evolution of Li Auto's VLA-based autonomous-driving architecture, scaling imitation learning, reinforcement learning, model size, and compute for safer and more efficient real-world driving.
Read launch coverageMach Mind
Released the Mach Mind-Pro and Mach Mind-Edge model family for cloud and on-device agentic intelligence, enabling native vehicle control, multimodal interaction, and always-on local perception.
Read model coverageResearch Interests
The current themes that define my research and engineering agenda.
Work Experience
Programs and roles that shaped my approach to applied AI systems.
Li Auto
Apr 2021 - Present
- Lead Li Auto's foundation-model and autonomous-driving teams, unifying Mach VLA, Mach Mind, 3D ViT, world models, reinforcement learning, data infrastructure, and on-vehicle deployment.
- Released the Mach VLA and Mach Mind-Pro / Mach Mind-Edge model series for embodied intelligence, with production integration into Livis vehicles and Li Auto's full-stack AI platform.
- Drive chip-model co-design and mass-production deployment around self-developed AI chips, vehicle operating systems, and domain controllers.
- Built Li Auto's autonomous-driving stack from Highway NoA and City NoA to E2E, VLM-assisted, and VLA-based architectures operating on hundreds of thousands of vehicles.
- Mentor a 100+ member organization across perception, planning, foundation models, simulation, data, and deployment.
- Launched Li Auto's overseas research hub, covering local strategy, budgeting, and talent acquisition.
- Bridge Silicon Valley innovation with Beijing execution through cross-border program reviews and roadmap alignment.
Baidu Apollo
Apr 2016 - Mar 2021
- Led the L4 prediction and pre-decision algorithms for robo-taxi pilots, improving motion forecasting in complex urban scenes.
- Shipped planning-and-control modules and deep-learning onboard components for autonomous fleets in Beijing and Guangzhou.
学术成果
基于 Google Scholar 的论文与引用快照
Top 10 引用论文
按 Google Scholar 引用量排序，更新时间见卡片上方。
Patents, Service & Community
Research service and technology transfer beyond the production stack.
Patents
20 granted or issued patents: 18 CN and 2 US across perception, planning, and HD mapping pipelines.
Reviewer
CVPR, ICCV, ECCV, NeurIPS, AAAI, IROS, and journals including TPAMI, T-ITS, and T-IV.
Community
Livis Day keynote speaker, organizer of the CVPR 2023 Autonomous Driving Workshop, and frequent speaker on VLA deployment in production.