About the job
About Us
At Physical Intelligence, we are pioneers in integrating general-purpose AI into the physical realm. Our team comprises dedicated engineers, scientists, roboticists, and innovators focused on creating advanced foundation models and learning algorithms that will empower the robots of today and the physically-actuated devices of tomorrow.
To achieve outstanding real-world performance, we prioritize ultra-low system latency, reliable sensor pipelines, and comprehensive engineering that ensures perception and control loops function seamlessly at real-time speeds.
As a Runtime Software Engineer, you will be at the forefront of developing low-latency, high-throughput systems that support our physical intelligence model. Your role will not involve designing ML models; instead, you will optimize the entire stack from the operating system to the camera pipeline and networking, ensuring flawless production execution. You will work closely with researchers, platform engineers, and robotics operators to identify performance bottlenecks and maximize system efficiency.
The Team
Our Runtime team is crucial in creating the foundational platform that our robots, sensors, and evaluation systems depend on. This team excels in Linux systems engineering, camera and sensor integration, robot actuator control, networking, real-time input/output, and performance optimization tools. They ensure that our machine learning models and control systems function within strict latency constraints and are resilient under real-world conditions.
Your Responsibilities
- Manage Real-Time Pipelines: Design and implement low-latency, high-reliability sensor and actuator pipelines utilizing Linux, drivers, and middleware.
- Enhance System Performance: Analyze and optimize computational efficiency across I/O, memory, scheduling, networking, and storage to satisfy real-time requirements and boost throughput.
- Develop OS-Level Features: Modify or expand Linux components, drivers, and scheduling mechanisms to ensure deterministic performance under load.
- Streaming & Video Systems: Create and refine real-time video streaming systems with precise frame timing and packet scheduling.
- Ensure Reliability & Debugging: Develop tools for profiling, tracing, and resolving timing challenges across distributed systems and hardware interfaces.
- Collaborate Across Functions: Partner with researchers, hardware engineers, and operational teams to drive system performance and reliability.
