About the job
Join Our Innovative Team
At OpenAI, our Consumer Products Research team is at the forefront of shaping the future of computing. We delve into cutting-edge modalities, interaction patterns, and system behaviors, engineering them into robust prototypes. The Neosensing team operates at the confluence of sensing technologies, edge algorithms, and systems engineering. We develop comprehensive software solutions that transform novel signals into reliable capabilities, including collection tools, integration protocols, and stable on-device loops that perform reliably in dynamic environments. We are passionate about software excellence and rapid iteration, emphasizing clean interfaces, debuggability, observability, and high performance even under strict device constraints.
Your Contribution
As a Software Engineer in our Consumer Products Research team, you will bridge the gap between algorithm development and implementable systems. Collaborating closely with algorithm engineers, you will convert prototypes into robust interfaces, dependable data pipelines, and optimized on-device solutions, with a sharp focus on performance, observability, and resilience against real-world challenges.
This role prioritizes software development, seeking a candidate who is passionate about writing high-quality code, takes pride in engineering craftsmanship, and is willing to dive deep into algorithmic intricacies to ensure seamless end-to-end functionality.
Work Environment
This position is based in San Francisco, CA, and follows a hybrid work model with four days in the office each week. Relocation assistance is available for new hires.
Key Responsibilities:
Develop and deploy pioneering production software for sensing algorithms, transforming algorithm prototypes into reliable end-to-end systems.
Manage and enhance critical components of the Python shipping pipeline, including integration surfaces, evaluation hooks, and performance quality safeguards.
Create embedded and on-device software within an RTOS environment (e.g., Zephyr) and implement models across various device runtimes and hardware accelerators.
Refine real-time on-device perception loops (e.g., detection/tracking pipelines) to ensure stability, low latency, and efficient use of power and memory.
Design and develop data collection and instrumentation tools that facilitate the introduction of new sensing modalities and expedite the process from prototype to dataset to model to device.
Collaborate cross-functionally with teams in algorithms, human data, and firmware/hardware to debug, profile, and enhance systems against real-world variability.

