Building in the New Spatial Intelligence Era | Spatial Frontier Club x Mission Robotics

About the Event SFC and Mission Robotics are bringing the community together for another evening of Spatial Intelligence in San Francisco! Over the past few months, the Spatial Intelligence landscape has been moving quickly. A new wave of models and tools is making capabilities across 3D generation, reconstruction, spatial reasoning, world modeling, and interaction with physical environments increasingly accessible. For researchers and builders across World Models, Robotics, 3D, Computer Vision, and Physical AI, this creates a new set of questions around what these capabilities actually unlock, how existing workflows and research directions may change, and where the hardest problems still remain. This gathering is designed for researchers, engineers, founders, and builders to share recent work, exchange ideas, and discuss what this shift means for the broader Spatial Intelligence ecosystem. The evening will combine short technical talks, open discussion, and networking. Come with a project to share, a question to explore, or simply an interest in what others are building. Who Should Attend Spatial IntelligenceWorld ModelsRobotics & Physical AIRobot Learning & Embodied AI3D Generative AIComputer Vision & Multimodal AIComputer GraphicsNeural Rendering & Gaussian SplattingSimulation & Digital TwinsXR & Spatial Computing We welcome researchers, engineers, technical artists, founders, product leaders, investors, students, and ecosystem builders whose work touches these areas. Event Format 5:30 PM – 6:15 PMArrival, networking, food & drinks 6:15 PM – 7:30 PMSFC & Mission Robotics welcome, speaker talks and technical sharing 7:30 PM – 8:00 PMOpen discussion and audience Q&A 8:00 PM – 9:30 PMCommunity mixer and networking Featured Speakers 1. Salman Shahid from Google DeepMind Salman Shahid is a Research Engineer at Google DeepMind, working on Embodied Reasoning. His research focuses on training robot policies with high dexterity while preserving the generalization capabilities of underlying models. Before joining Google DeepMind, he conducted independent research on Stanford’s BEHAVIOR robotics challenge, fine-tuning Physical Intelligence’s π0.5 VLA model. His work found that proprioceptive dropout can significantly improve policy robustness, while receding-horizon control outperformed temporal ensemble methods, improving door navigation success from 30% to 100%. 2. Hokin Deng from Carnegie Mellon University Hokin Deng is a researcher working on video reasoning AI, studying how machines can understand and predict the physical world from video. He leads VBVR, a large-scale video model benchmark spanning 56 authors across 32 universities, which reached #1 on Hugging Face and was named Paper of the Month, alongside publications at ICML. He previously conducted research in cognitive science and neuroscience at Johns Hopkins, Harvard and Carnegie Mellon. Paper: arxiv.org/abs/2608.26105 3. Jake Fitzgerald Creator of text-to-cad Jake has spent the past six months exploring agentic hardware design at South Park Commons. He created text-to-cad, an open-source project with 16K GitHub stars and 200K downloads, equipping AI agents with tools for 3D modeling, simulation, manufacturability checks, and robot planning. Using these tools, he designed and built a six-degree-of-freedom robot arm—without any prior experience building robots. In this talk, he’ll share the possibilities and practical challenges of using LLMs to design and build hardware. Previously, Jake co-founded Derive Labs, a crypto options exchange, and worked as a software engineer at WhatsApp. texttocad.dev/ Call for Speakers We’re currently inviting speakers to share recent research, technical demos, and insights from their work. We’re especially interested in work exploring how recent advances in spatial models and tooling are changing what’s possible across World Models, Robotics, 3D, Spatial Reasoning, Simulation, and Physical AI.…