Principal AI SoC Runtime Software Architect

We are looking for a Principal AI SoC Runtime Software Architect with 10+ years of experience to own the runtime at the center of Velaura's ultra-low-power AI SoC for robotics and physical AI. You'll be a hands-on, senior technical leader with deep embedded or on-chip runtime experience, ideally from imaging, multimedia or AI inference pipelines on silicon. This is a key early hire: you'll define the architecture, then build and lead a team of engineers around it.

What will you be doing?

Architect and build the embedded runtime that moves data from sensors (camera, audio, IMU) through image processing, preprocessing, NPU inference and postprocessing, then up to robotics applications.

Coordinate execution across heterogeneous engines (CPUs, ISPs, DSPs and the NPU), getting the most out of DMA, caching, shared buffers and synchronization for predictable, low-latency pipelines.

Define the interfaces between the runtime, Linux kernel drivers, firmware, the compiler and the Velaura SDK, including stable APIs, observability and fault recovery.

Debug and tune end-to-end performance problems that cross software and hardware layers, measuring latency, throughput, bandwidth and power.

Lead the work across runtime, kernel, driver, multimedia and SDK engineers, and help hire the team that grows around this role.