Perception Engineer
Production Computer Vision · NVIDIA Edge Systems
San Francisco, CA | 4 days onsite, 1 day remote
$220,000–$270,000 base + competitive equity | Full-time | Relocation available
Build perception systems that bring advances in autonomous robotics and self-driving technology to residential security.
Our client is an early-stage startup with $18M raised and a team of approximately 25. Its home security platform is already deployed in the field through active beta programs. The team is looking for a hands-on Perception Engineer to take ownership of the perception stack and improve how the system detects, tracks, and understands activity around a home.
What You’ll Own
You’ll work across models, video pipelines, and edge deployment, turning computer vision capabilities into reliable features running on NVIDIA hardware.
-
Scale the perception pipeline: Re-architect the existing service on NVIDIA Thor to support more camera streams and improve stability.
-
Improve real-time performance: Evaluate and deploy faster, more capable models within edge compute constraints.
-
Add depth and tracking capabilities: Bring monocular depth estimation into production to support 3D tracking of people and vehicles.
-
Optimize the NVIDIA stack: Work with DeepStream, JetPack, and TensorRT to improve inference and video processing.
-
Ship across the system: Partner with device and systems engineers to deliver improvements to deployed hardware.
This is an individual contributor role with substantial technical ownership.
What You Bring
-
Approximately 3–7 years of relevant experience, including ownership of production perception or computer vision systems.
-
Strong C++ and Python skills and experience writing production software.
-
Hands-on experience with NVIDIA DeepStream, JetPack, and TensorRT, ideally deploying on Jetson Thor or Orin.
-
Production experience with GStreamer and/or FFmpeg video pipelines.
-
Applied computer vision and deep learning experience, including model deployment, PyTorch, multi-object tracking, and multi-camera association.
-
Experience making architecture and performance decisions for systems deployed on edge hardware.
-
A background in a safety-critical field such as robotics, autonomous vehicles, aerospace, medical devices, or physical security.
-
A BS or MS in Computer Science, or equivalent experience.
The strongest candidates can explain what they built, the tradeoffs they made, and how their systems performed after deployment.
Compensation & Benefits
-
$220,000–$270,000 base salary, with flexibility for exceptional candidates
-
Competitive seed-stage equity
-
Relocation assistance
-
Health insurance, 401(k), commuter benefits, and fertility benefits
Candidates on OPT or seeking an H-1B transfer may be considered. New visa sponsorship is not available.
Interview Process
-
Hiring manager conversation — 30 minutes, virtual: Your experience, ownership approach, and interest in an early-stage team.
-
Technical conversation — 45 minutes, virtual: A project deep dive and live perception/ML problem-solving, rather than LeetCode-style exercises.
-
Onsite interviews — San Francisco: Four sessions covering perception and ML, device and video pipelines, system design and software engineering, and collaboration and leadership.