Senior Perception Engineer

5V Tech; California (MO); On-site; USD 120,000 - 190,000; Full time

Benefits offered by this job

Competitive salary

Meaningful equity

Health, dental and vision benefits

Job summary

5V Tech is seeking a software engineer to build and ship the perception stack for real-time robot understanding. You’ll own data, training, evaluation and deployment on production hardware, enabling accurate detection, tracking and 6D pose estimation on robotic systems.

You’ll fuse camera, LiDAR, depth and IMU data to create a single reliable world view, pushing models from training to on-robot deployment within latency constraints of onboard hardware.

Qualifications

• 3+ years building perception, computer vision or SLAM deployed on hardware (robots, drones, autonomous vehicles).
• Strong C++ and/or Python, with PyTorch experience.
• Solid 3D geometry and camera model fundamentals.

Responsibilities

• Build real-time detection, segmentation and tracking for robot camera feeds.
• Develop and improve SLAM and visual-inertial odometry for robots navigating busy environments.
• Deliver 3D perception, including depth, point clouds and 6D pose estimation for manipulation and grasping.
• Fuse camera, LiDAR, depth and IMU data into a single reliable view of the world.
• Ensure models run within latency/compute limits of onboard hardware.
• Identify and fix failure cases using data and evaluation pipelines.

Skills

C++

Python

PyTorch

Tools

PyTorch

Job description

Have you shipped perception on a real robot?

Plenty of engineers have trained vision models. Far fewer have had one running on a robot that has to act on what it sees, in real time, around real people. If that's you, this is worth five minutes.

About the company

Our client is a venture-backed US robotics company with robots already deployed commercially. Their perception team owns how the robot sees and understands the world, and they're growing it to support the next phase of deployment and scale.

The role

You’ll build and ship the perception stack that lets the robot detect what's around it, know exactly where it is, and understand objects precisely enough to act on them. You’ll own your work end to end, from data and training through to evaluation and deployment on the robot.

What you’ll do

• Build real-time detection, segmentation and tracking for robot camera feeds
• Develop and improve SLAM and visual-inertial odometry for robots navigating busy, changing environments
• Deliver 3D perception, including depth, point clouds and 6D pose estimation for manipulation and grasping
• Fuse camera, LiDAR, depth and IMU data into one reliable view of the world
• Get models running within the latency and compute limits of the robot’s onboard hardware
• Find and fix failure cases in the field using data and evaluation pipelines

What you’ll bring

• 3+ years building perception, computer vision or SLAM that has been deployed on physical hardware: robots, drones or autonomous vehicles
• Strong C++ and/or Python, with PyTorch
• Solid 3D geometry and camera model fundamentals
• Experience taking a model from training to real-time deployment, and supporting it once it’s live

Nice to have

• 6D pose estimation, grasp detection or bin picking
• Multi-camera calibration or sensor fusion
• Autonomous vehicle or AR/VR perception experience (the skills transfer well, and these teams are open to it)

This probably isn’t the right fit if...

Your computer vision experience is mainly in 2D image classification, retail analytics, content moderation or medical imaging, with no hardware deployment. If you’re keen to move into robotics from one of those areas, still get in touch, as there may be a better-matched route.

What’s on offer

• Competitive base salary and meaningful equity
• Full health, dental and vision benefits
• Your work running on real robots in production, not just in simulation
• A small, highly technical team where each perception hire shapes the product