Peripheral

Peripheral is a Toronto company building volumetric capture for live sport. It was founded in 2024 by Kelvin Cui and Mustafa Khan, who both worked on the University of Toronto's driverless car team, Khan as a researcher at Huawei and Cui as a software engineer at Tesla working on chassis systems. The company describes its work as spatial intelligence for media, starting with live sports and entertainment, and frames the same models as groundwork for robotic perception and embodied intelligence.

The headline claim is a reduction in rig size. Volumetric capture of a venue has typically required more than 100 synchronized cameras, and Peripheral puts its own requirement at as few as 32, which it aims at install cost and operational overhead. It builds its own sensor stack, which it likens to the depth capturing sensors used on self driving cars, while working with off the shelf cameras. The company says the packaging of that stack, rather than the cameras themselves, is what lets it scale from small practice enclosures to full soccer and football stadiums.

The platform gives broadcasters and fans control of the viewpoint, including locking onto a single player or freezing a moment and moving around it. The same capture yields biomechanical data, tracking individual joints down to finger movement and measuring knee and ankle flexion for coaching use. Peripheral established a biomechanics basketball shooting lab in Toronto.

Peripheral names no scene representation on any product facing surface. Its hiring material does. The listing for a Senior Machine Learning Engineer in reconstruction requires deep expertise in multi view geometry, structure from motion, NeRFs, 3D and 4D gaussian splatting, and modern feed forward 3D reconstruction, and sets the role's agenda as real time feed forward photorealistic 3D and 4D reconstruction. That is hiring material rather than a product claim, and it is the only public statement connecting the company to these methods.

Peripheral raised a 3.6 million dollar seed round led by Khosla Ventures. Published investor lists differ between the company's own job postings and press coverage, and no single list should be treated as complete. The company had 10 engineers at the time of the round and lists five open roles, all onsite in Toronto. It has published no product page, no documentation, no demo and no dated announcement of its own, and its public website is one sentence long.