About
Check my website - https://wenhao.pub
Activity
1K followers
-
Wenhao Ding shared thisWe’re hiring research interns to work on foundation models and reasoning for Physical AI, spanning both humanoid robotics and autonomous driving. This is not limited to summer internships - start dates are flexible throughout 2027. Please apply if you’re interested!Wenhao Ding shared thisWe are hiring! The Autonomous Systems and Physical AI Research (ASPIRE: https://lnkd.in/gK4V8aSy) group at NVIDIA is looking for talented PhD Research Interns to join us in advancing the frontiers of autonomous systems and Physical AI. We work across a broad range of research areas, including reasoning models, generative simulation, agentic AI workflows, and Physical AI safety, with applications spanning autonomous vehicles and a broad range of Physical AI systems. Interested in pushing the limits of what’s possible? Apply now: https://lnkd.in/g-sVuXFG NVIDIA DRIVE NVIDIA Robotics NVIDIA AIPhD Research Intern, Autonomous Systems and Physical AI Research - 2027PhD Research Intern, Autonomous Systems and Physical AI Research - 2027
-
Wenhao Ding reposted thisWenhao Ding reposted thisWe have just released #Alpamayo 2 Super — NVIDIA’s frontier open reasoning model for #autonomous #vehicles. NVIDIA Alpamayo 2 Super is an open 34-billion-parameter #reasoning vision-language-action (#VLA) model designed to accelerate autonomous vehicle (AV) development. It combines the 32-billion-parameter NVIDIA #Cosmos 3 Super Reasoner with a 2-billion-parameter diffusion-based Action Expert and is post-trained with #reinforcement #learning. Two aspects make Alpamayo 2 Super particularly exciting: 1. Open and commercially deployable: Alpamayo 2 Super is available on Hugging Face under OpenMDW-1.1, the Linux Foundation’s permissive license for open AI model distribution. The OpenMDW license is now being applied across the entire Alpamayo model family, enabling developers to deploy these models commercially without requiring additional permissions. 2. A multi-task foundation model for autonomous driving: Alpamayo 2 Super produces five tightly coupled outputs: - A trajectory describing the vehicle’s planned path. - A chain-of-causation (CoC) trace explaining the reasoning behind the decision, achieving benchmark-leading reasoning performance at frontier scale. - A meta-action (e.g., yield, lane change, stop) capturing the model’s intent. - Reasoning auto-labels that generate CoC annotations for training and validation data. - Visual question answering responses with 2D visual grounding, linking answers to specific regions in camera images. These multi-task capabilities enable developers to leverage a single foundation model across more of the development process, simplifying tooling and accelerating iteration. Resources: 🔹 Blog: https://lnkd.in/g-kT69s4 🔹 Technical blog: https://lnkd.in/gNe5rdQZ 🔹 Hugging Face blog: https://lnkd.in/gMYJS7WA 🔹 Model weights: https://lnkd.in/g3qYyQNU 🔹 Inference code: https://lnkd.in/gTZ5ybJP 🔹 Video: https://lnkd.in/gtjwmZFs 🔹 Interview: https://lnkd.in/guZ9REbc As Jensen Huang has emphasized, open models help advance safety and security. We hope Alpamayo 2 Super will contribute to this vision by enabling researchers and developers around the world to build, experiment, and innovate in autonomous driving. We are excited to see what the community builds with Alpamayo 2 Super. NVIDIA DRIVE NVIDIA AI
-
Wenhao Ding reposted thisWenhao Ding reposted thisToday, we’re launching Alpamayo 2 Super, our frontier open reasoning model for autonomous vehicles. Beyond seeing, Alpamayo understands and reasons through the complex world - thinks before it acts. It’s a powerful backbone for robotaxis, trucks, shuttles, delivery vans, tractors and the long tail of mobile robots—billions of autonomous machines someday. We’re releasing it for commercial use under OpenMDW-1.1 so teams can inspect it, fine-tune it and deploy it—open models advance safety and security. The next wave of AI is robotics—and it starts with autonomous vehicles. Great work, Alpamayo team! https://lnkd.in/g7urPEjp
-
Wenhao Ding shared thisWenhao Ding shared this🚗🏆 Breaking news: NVIDIA #Alpamayo open platform has been named the winner of a COMPUTEX TAIPEI 2026 Best Choice Award in the Vehicle Technology & Smart Cockpit category, recognizing Alpamayo as one of the year’s major breakthroughs in automotive and Physical AI technology! Announcements: — https://lnkd.in/gvQAgDME — https://lnkd.in/gFy68EwY I’m incredibly proud of the team behind the Alpamayo open platform — Wenjie Luo Yan Wang Boris Ivanovic Maximilian Igl Michael Watson Edward Schmerling and the entire NVIDIA Autonomous Vehicle Research Group — and deeply grateful for the contributions from the NVIDIA AV production team Xinzhou Wu Sarah Tariq and many other researchers and developers across NVIDIA. This achievement was truly a collective team effort.👏 To get started with Alpamayo: — https://lnkd.in/eHJ4r67Q — https://lnkd.in/g-K7Pgza And stay tuned — we’ll have several exciting announcements in the coming weeks. NVIDIA DRIVE NVIDIA AIExpanding the Alpamayo Open Platform for Developing Reasoning AVs Across Models, Data, and SimulationExpanding the Alpamayo Open Platform for Developing Reasoning AVs Across Models, Data, and Simulation
-
Wenhao Ding reposted thisWenhao Ding reposted thisJensen today announced Alpamayo 1.5 at #NVIDIAGTC! #Alpamayo 1.5 is a major update to Alpamayo 1—NVIDIA’s open 10B-parameter chain-of-thought reasoning VLA model, first introduced at #CES. Built on the #Cosmos-Reason2 VLM backbone and post-trained with RL, it adds support for navigation guidance, flexible multi-camera setups, configurable camera parameters, and user question answering. The result is an interactive, steerable reasoning engine for the AV community. We’re also releasing post-training scripts to help researchers and developers adapt the model. Additionally, we’ve significantly expanded the Alpamayo open platform across data and simulation, including releasing highly requested reasoning labels for the PhysicalAI Autonomous Vehicles dataset (https://lnkd.in/g3As7huw), as well as our chain-of-causation auto-labeling pipeline. 🔎 Learn more about Alpamayo 1.5 and the latest extensions to the Alpamayo open platform: https://lnkd.in/g-K7Pgza (please note that most of the links will become active in the next few days.) Happy building—and stay tuned for more in the coming months! NVIDIA DRIVE NVIDIA AI
-
Wenhao Ding reposted thisWenhao Ding reposted thisWhat does it take to build autonomous vehicles that can reason about the world they drive in? Tomorrow at #NVIDIAGTC, Patrick Langechuan Liu and I will take a deep dive into the Alpamayo reasoning model family—a family of reasoning-based vision–language–action (VLA) models that form a core component of the Alpamayo open platform (https://lnkd.in/eHJ4r67Q). We’ll cover three main topics: - How reasoning-based VLA models like Alpamayo 1 are designed and built - What it takes to bring Alpamayo 1 to production, including some of our latest results - Several exciting announcements about the expansion of the Alpamayo open platform If you’re working on autonomous driving, robotics, or foundation models for physical AI, this session will offer a look at where the field is heading. Session details: 📅 Monday, Mar 16 | 3:00 PM PDT 📍 #NVIDIAGTC 2026 🔗 https://lnkd.in/g5sf8Rm6 Looking forward to seeing many of you there. NVIDIA DRIVE NVIDIA AI
-
Wenhao Ding shared thisWenhao Ding shared this🚗 How does an autonomous vehicle think in real time? Tomorrow, NVIDIA AV researchers go live to unpack Alpamayo 1—a chain-of-thought reasoning VLA model that makes the logic behind every driving decision explicit. Hear from Marco Pavone, Yan Wang, Yurong You, and Wenhao Ding as they walk through how reasoning-driven VLA models are shaping the future of autonomous driving. 🕘 Tomorrow | Wednesday, Feb 11 | 9:00–10:00am PST ▶️ Watch live on YouTube: https://nvda.ws/4khpE0Q
-
Wenhao Ding reposted thisWenhao Ding reposted thisMore on #reasoning in Vision-Language-Action (#VLA) models --- Traditional VLA models decide what action to take by decomposing complex situations into their most salient factors. But reasoning models can do much more. When viewed as implicit world models operating in a semantic space, they can be used counterfactually—exploring multiple “what if” scenarios before acting. In our recent paper, Counterfactual VLA (CF-VLA, https://lnkd.in/gGgszkib), we show that counterfactual reasoning consistently improves trajectory accuracy, safety, and reasoning quality. Key contributions: - Self-reflective counterfactual reasoning: CF-VLA reflects on predicted meta-actions, anticipates consequences, and revises plans before execution—enabling causal self-correction. - Automated data pipeline: A novel data pipeline generates counterfactual data, forming a self-improving loop for reasoning and action. - Adaptive thinking in autonomous driving: CF-VLA focuses reasoning on the most challenging scenarios, improving performance while keeping test-time computation efficient. Paper: https://lnkd.in/gGgszkib #AI #Robotics #VisionLanguageAction #AutonomousSystems #MachineLearning #CounterfactualReasoning NVIDIA AI NVIDIA DRIVE
-
Wenhao Ding reposted thisWenhao Ding reposted this🚗 Imitation learning is everywhere—but is it enough? So far, imitation learning—most commonly via behavior cloning (BC)—remains the go-to approach for training real-world autonomous vehicle (AV) driving policies. Yet BC operates in an open-loop (OL) fashion, overlooking the critical interdependence among inputs, outputs, and future states that comes with closed-loop (CL) operation. The result? The notorious—but often overlooked—OL–CL gap ⚠️ To address this challenge and encourage broader adoption of CL techniques, we’ve just published a survey (https://lnkd.in/g_tKSZ_X) presenting a comprehensive taxonomy of closed-loop training methods for end-to-end driving. Our framework organizes approaches along three key axes: - Action generation - Environment response generation - Training objectives 💡 Bottom line: enabling technologies—like neural rendering, generative world models, and scalable RL—have now matured, making closed-loop AV training ready for wide-scale adoption. We’d love to hear your thoughts—drop a comment and join the discussion! 💬 And as a reminder, we are hiring for full-time research scientist and research engineer positions: 🔹 [Sr.] Research Scientist: https://lnkd.in/dbsmr_SH 🔹 [Sr.] Research Engineer: https://lnkd.in/gSEj2Hbp NVIDIA DRIVE NVIDIA AI
-
Wenhao Ding liked thisWenhao Ding liked thisFollowing up on my previous post about PhD Research Intern opportunities, the Autonomous Systems and Physical AI Research (ASPIRE, https://lnkd.in/gK4V8aSy) group at NVIDIA is also hiring for full-time positions! Priority will be given to candidates who can start by the end of January 2027. Open positions: - (Senior) Research Scientist — Apply here: https://lnkd.in/ev2tViwH - (Senior) Research Engineer — Apply here: https://lnkd.in/eYwPX7GC We work across #reasoning models, #generative simulation, #agentic #AI workflows, and #Physical AI safety, with applications spanning autonomous vehicles and a broad range of Physical AI systems. If you’re interested in joining us, we’d love to hear from you! NVIDIA DRIVE NVIDIA Robotics NVIDIA AI
-
Wenhao Ding liked thisI am thrilled that Hyundai Motor Group is building their autonomous driving roadmap on NVIDIA. Their dual-track strategy brings Nvidia Hyperion platform and NVIDIA DRIVE solution with Alpamayo into Hyundai's software-defined vehicle architecture, targeting Level 2+ production vehicles in H1 2028 and Level 2++ in H2 2028. Really looking forward to the close partnership and collaboration between NVIDIA and Hyundai Motor Company, also between my team and Minwoo P.'s team.Wenhao Ding liked thisHyundai Motor Group unveiled its autonomous driving development roadmap at HMG Autonomous Driving Media Day, including the first public footage of Atria AI navigating complex Seoul traffic. At the core is a Data Flywheel connecting real-world data collection, AI training, validation and deployment, supporting a dual-track strategy that combines NVIDIA-based production technologies with the development of proprietary Atria AI. Watch the footage ▶ https://lnkd.in/gze23NgZ #HyundaiMotorGroup #42dot #AtriaAI #DataFlywheel #AutonomousDriving #VLAHyundai Motor Group Accelerates Autonomous Driving Innovation with AI-Powered Data FlywheelHyundai Motor Group Accelerates Autonomous Driving Innovation with AI-Powered Data Flywheel
-
Wenhao Ding liked thisWenhao Ding liked thisI’m incredibly excited to introduce Veeda AI, which I co-founded with my incredible longtime collaborators Zan Gojcic and Huan Ling. We started Veeda because we believe robotics will reshape the world—changing how we move people and goods, how we manufacture and build, and how we operate in the physical world. We also firmly believe that the scaling moment for Physical AI will come from robots learning through interaction with the world. Just like humans learn from interaction with their environments by trying, failing and trying again, until we succeed. And just like the capabilities of LLMs were truly unlocked when they started training in interactive environments and not only on vast amounts of internet text. For Physical AI, the central challenge is making this kind of interactive learning possible at scale. The real world is simply not a practical training ground for robots to learn through trial and error. It is unsafe, expensive, and time-consuming. To scale interactive learning, robots will need to learn in simulated reality. At Veeda, our sole mission is to build simulated reality for Physical AI. Our conviction is that this will become the critical infrastructure layer for all areas of robotics. As envisioned in pop culture, this means building “the Matrix” for Physical AI, where, through a virtual embodiment, robot intelligence can interact with an entirely virtual world in a million possible ways, learning from its own mistakes. We believe that World Models are the foundational technology that can make this possible. These generative models learn from massive amounts of sensor and physical-world data to reach the quality, diversity, and physical realism of the simulated reality that robots require. I could not be more excited to take on this challenge and build this transformative technology alongside an extraordinary team of engineers and researchers who have spent years working on some of the hardest problems across simulation, generative AI, 3D, robotics, and embodied intelligence. Together with our investors at Khosla Ventures and Radical Ventures, we intend to push the frontier of this technology and build the foundational infrastructure for interactive robotics learning and evaluation at scale. Come build this future with us. And to roboticists, help us understand what you need, we’re building this for you. veeda.ai
-
Wenhao Ding liked thisWenhao Ding liked thisToday, we’re launching Alpamayo 2 Super, our frontier open reasoning model for autonomous vehicles. Beyond seeing, Alpamayo understands and reasons through the complex world - thinks before it acts. It’s a powerful backbone for robotaxis, trucks, shuttles, delivery vans, tractors and the long tail of mobile robots—billions of autonomous machines someday. We’re releasing it for commercial use under OpenMDW-1.1 so teams can inspect it, fine-tune it and deploy it—open models advance safety and security. The next wave of AI is robotics—and it starts with autonomous vehicles. Great work, Alpamayo team! https://lnkd.in/g7urPEjp
-
Wenhao Ding liked thisWenhao Ding liked thisHappy to share that my latest approach now ranks #1 on the KITScenes LongTail Challenge leaderboard (CVPR Workshop on Autonomous Driving). During the official competition, my team finished close to the podium. After the challenge ended, I continued working on the problem, and the latest version of the approach now achieves an MMS score of 5.54, improving the previous best score of 5.21. The approach combines Alpamayo 1.5 for trajectory generation with SAM3 for automatic scene segmentation, enabling the system to identify traversable areas and refine trajectory predictions based on scene geometry and environmental constraints. Many thanks to the challenge organizers and the authors of the KITScenes dataset for creating such an interesting benchmark and making it available to the research community. Leaderboard: https://lnkd.in/e-PQsKFa #CVPR #AutonomousDriving #ComputerVision #DeepLearning #TrajectoryPrediction #AI
-
Wenhao Ding liked thisWenhao Ding liked thisI’m excited to share that I will join NC State Mechanical and Aerospace Engineering as a 𝐭𝐞𝐧𝐮𝐫𝐞-𝐭𝐫𝐚𝐜𝐤 𝐀𝐬𝐬𝐢𝐬𝐭𝐚𝐧𝐭 𝐏𝐫𝐨𝐟𝐞𝐬𝐬𝐨𝐫 in Fall 2026! 🐺🎉 At NC State, I will launch the 𝑷𝑨𝑹𝑰𝑺 (𝑷𝒉𝒚𝒔𝒊𝒄𝒂𝒍 𝑨𝑰, 𝑹𝒐𝒃𝒐𝒕𝒊𝒄𝒔, 𝒂𝒏𝒅 𝑰𝒏𝒕𝒆𝒍𝒍𝒊𝒈𝒆𝒏𝒕 𝑺𝒚𝒔𝒕𝒆𝒎𝒔) 𝑳𝒂𝒃 🌍🤖 with a mission to advance intelligent systems that can perceive, reason, learn, and act in the physical world. The PARIS Lab will pursue three interconnected research directions: 🧠 𝐄𝐦𝐛𝐨𝐝𝐢𝐞𝐝 𝐢𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐜𝐞 𝐚𝐧𝐝 𝐏𝐡𝐲𝐬𝐢𝐜𝐚𝐥 𝐀𝐈: developing AI models and algorithms that enable machines to understand complex environments, reason about actions, and make reliable decisions in real-world settings. 🤖 𝐅𝐨𝐮𝐧𝐝𝐚𝐭𝐢𝐨𝐧-𝐦𝐨𝐝𝐞𝐥-𝐝𝐫𝐢𝐯𝐞𝐧 𝐫𝐨𝐛𝐨𝐭𝐢𝐜𝐬: advancing robot navigation and manipulation through vision-language-action models, world models, and generalizable robot learning frameworks. 🤝 𝐇𝐮𝐦𝐚𝐧-𝐜𝐞𝐧𝐭𝐞𝐫𝐞𝐝 𝐚𝐧𝐝 𝐚𝐝𝐚𝐩𝐭𝐢𝐯𝐞 𝐀𝐈 𝐬𝐲𝐬𝐭𝐞𝐦𝐬: creating AI and robotic systems that can collaborate naturally with people, continuously learn from interaction, and adapt safely to new tasks and environments. 𝐈 𝐚𝐦 𝐫𝐞𝐜𝐫𝐮𝐢𝐭𝐢𝐧𝐠 𝐏𝐡.𝐃. 𝐬𝐭𝐮𝐝𝐞𝐧𝐭𝐬 𝐟𝐨𝐫 𝐒𝐩𝐫𝐢𝐧𝐠 𝟐𝟎𝟐𝟕 𝐚𝐧𝐝 𝐅𝐚𝐥𝐥 𝟐𝟎𝟐𝟕. Prospective students interested in physical AI, robotics, robot learning, foundation models, and human-AI collaboration are very welcome to reach out. If you are interested in working with me, please reach out without hesitation and send your 𝐂𝐕 𝐚𝐧𝐝 𝐚 𝐛𝐫𝐢𝐞𝐟 𝐫𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐬𝐭𝐚𝐭𝐞𝐦𝐞𝐧𝐭 to zhuang46@ncsu.edu I also warmly welcome collaborations with researchers, students, and industry partners who share similar interests. I am deeply grateful to Jiaqi Ma at UCLA for his postdoctoral advising, mentorship, and support throughout this journey. I would also like to sincerely thank my Ph.D. supervisor, Chen Lv, for his guidance, encouragement, and support during my doctoral training and beyond. #NCState #PhysicalAI #Robotics #ArtificialIntelligence #RobotLearning #EmbodiedAI #FoundationModels #HumanAIInteraction #PhDRecruiting #AcademicJobs
Experience
Education
Other similar profiles
Explore more posts
-
Jigar Halani
Deploy Kimi-K2-Instruct, a frontier mixture-of-experts model, on Lambda using vLLM in four steps: 1️⃣ Spin up an NVIDIA Blackwell instance on Lambda. 2️⃣ Start a vLLM deployment server with a single setup block. 3️⃣ Run a reproducible benchmark against that server. 4️⃣ Share the exact model and image references to reproduce. Read the technical blog to get started 🔗
-
Martin Kielski
"AI as a Compiler" takes a different route to GPU optimization: give an LLM a Triton kernel and let it generate PTX directly, skipping Triton's conventional lowering pipeline. Across Ada, Hopper and Blackwell, the generated kernels reach 0.83x-3.34x the performance of autotuned Triton; BitDelta reaches 3.34x, FlashAttention 1.37x, while several simpler and memory-bound kernels stay close to Triton. The generated PTX uses architecture-specific mechanisms such as WGMMA, TMA, mbarrier and Blackwell's tcgen05, but verification remains a constraint because some of the optimizations responsible for higher performance fall outside the verifier's supported semantics. This points to a practical model for GPU software: keep the high-level kernel as the specification, search for architecture-specific low-level implementations offline, test and profile them, then reuse the validated result AI as a Compiler: Compiling Triton kernels without the Triton compiler https://lnkd.in/dkZ3KzUH #LLM #GPU #Triton #PTX #CUDA #AICompiler #Compiler #GPUProgramming #LLMInference #Inference #KernelOptimization #TensorCores #NVIDIA #Hopper #Blackwell #FlashAttention #MachineLearningSystems
-
Joe Pierce
Our NVIDIA Blackwell set a high bar in the latest results of SemiAnalysis_ InferenceMAX benchmarks. This new open source initiative provides a comprehensive methodology to evaluate inference hardware and software performance. Here are 5 key benchmark takeaways that demonstrate how NVIDIA GPUs deliver the highest inference performance across all workloads 👇 📈 DeepSeek-R1 results showed that NVIDIA Blackwell GB200 leads across the board and demonstrates a 15x performance and revenue improvement over NVIDIA Hopper H200. 📈 gpt-oss 1K/1K can achieve a max throughput of 60,000 TPS per GPU, max interactivity of 1,000 TPS/user and 5x performance improvement in just 2 months since gpt-oss launch. 📈 On both gpt-oss-120b and Llama 3.3 70B, NVIDIA Blackwell B200 achieves up to 4x performance improvement over NVIDIA Hopper H200. 📈 Llama 3.3 70B results show that NVIDIA Blackwell B200 NVFP4 leads the pack across all ISL/OSL configurations and latency targets. 📈 NVIDIA Dynamo disaggregated inference and TensorRT-LLM wide Expert Parallel combine with NVIDIA Blackwell to unleash the full performance of MoE models. Technical Deep Dive ➡️ https://bit.ly/4nSAxHH
-
Kosuke Tatsumura
Our paper, “Enhancing vehicle-mountable multiple object tracking systems with embeddable Ising machines,” has been published in Nature Communications: https://lnkd.in/gAnxyN4W In this work, we integrate an embeddable Ising machine—based on a quantum-inspired algorithm called simulated bifurcation—into a vehicle-mountable platform to advance real-time information processing for mobile vehicles and robots. The Ising machine simulates the dynamics of a quantum-inspired nonlinear oscillator network to find high-quality approximate solutions in cyberspace, and can be implemented on an FPGA. This integration enables the system to tackle computationally hard combinatorial optimization problems for autonomous control, including multiple object tracking and simultaneous localization and mapping.
2 Comments -
Ussama Naal
If you are in the Bay Area around the coming NVIDIA GTC event consider joining us at “ROS by the Bay” special event as me and the CEO of Ouster will share some details about the work that Ouster is doing to enable Physical AI and its adoption. We’ll also go over the recent acquisition of StereoLabs and how that fits into our roadmap. We’ll have some demos and be joined by many of other Ouster experts in SW and HW and Optics. Looking forward to see you there! #ouster #stereolabs #phsysicalai #rosbythebay #gtc #ros #robotics
-
Daniela Sharifi
I work closely with several leading autonomy and robotics companies in the Bay Area 🚗🤖 and regularly connect with engineers across planning, perception, simulation, and robotics systems. One trend I’m seeing across many teams: strong systems engineers are increasingly in demand. The stack (often C++ / Python) matters, but what matters more is deep domain experience in robotics or autonomous systems, particularly around safety, evaluation, validation, and fault handling. If you’re working in this space or curious about what’s happening across AV and robotics, feel free to reach out. #robotics #autonomoussystems #physicalAI
4 Comments
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content