Software Engineer, On-Device AI

Nothing Technology•San Francisco, CA

23h•$164,000 - $248,000

About The Position

Nothing is building an operating-system-level foundation for AI that runs across the devices people carry and can complete useful tasks. The Software Engineer, On-Device AI will build system services that keep small models running reliably on real hardware, along with APIs and abstractions that connect model capabilities to platform and app layers. The role involves end-to-end ownership of open-ended systems problems, balancing performance, latency, battery, thermal, privacy, and hardware constraints. The position is based in San Francisco, with remote work considered for the right person and regular in-person time with the San Francisco team. Nothing offers excellent benefits and competitive compensation. The stated base salary range is $164,000–$248,000 USD.

Requirements

Experience owning complex systems work from architecture through hands-on implementation and debugging.
Strong systems fundamentals, including concurrency, resource management, API design, and performance trade-offs.
Experience with on-device inference, inference runtimes, device software, or another resource-constrained platform; alternatively, demonstrated ability to learn quickly across these areas.
Ability to make progress without a specification, explain trade-offs clearly, and change direction based on data.
Willingness to question familiar architectures rather than simply copy them.
Strong programming skills and flexibility to use languages suited to the problem, such as Rust, Python, and JavaScript.

Responsibilities

Build core system services, abstractions, and APIs for a platform others can build on.
Run small models reliably on real hardware, managing memory, scheduling, and resources within tight latency, battery, and thermal budgets.
Connect model capabilities to platform and app layers so they can complete real tasks end to end.
Determine where tasks should run—on a wearable, phone, computer, or in the cloud—based on the task, privacy, and available hardware.
Prototype quickly, profile performance on real hardware across CPUs, GPUs, and NPUs, and ship solutions that work.
Work closely with software, model, and hardware engineers.