Perception Engineer

Perception Engineer – Zurich (Hybrid/Remote)

Robotics AI | Computer Vision | Edge Deployment

Most perception roles are about improving model performance in controlled conditions.

This one is about building perception systems that work in the real world.

I’m working with a fast-growing robotics AI company that is building advanced perception and autonomy technology for physical environments where reliability, latency and deployment quality genuinely matter.

They are looking for a Perception Engineer to take ownership of core computer vision systems across detection, tracking, re-identification, scene understanding and edge deployment.

This is not a narrow research role, and it is not a role where you hand models over and hope they work later.

You will be building systems that need to run live, handle messy real-world data, operate across multiple sensor streams, and continue performing outside the lab.

The role

You will work on the perception stack behind a real-world robotics AI platform, helping turn modern computer vision models into reliable deployed systems.

The work will sit across model development, optimisation, deployment and system-level thinking. You will be close to the product, close to the engineering team, and close to the practical challenges that come with deploying AI into physical environments.

This would suit someone who enjoys building useful systems, not just training models.

What you will be working on

• Building and improving perception pipelines for real-world robotic and sensor-based systems
• Developing detection, tracking and re-identification models for complex visual environments
• Working with high-throughput video and multi-stream perception data
• Applying modern computer vision and vision-language techniques to scene understanding
• Optimising models for latency, robustness and edge deployment
• Taking models from prototype stage into production-quality systems
• Working closely with autonomy, robotics, software and infrastructure engineers
• Debugging real-world failure modes and improving system reliability over time

What they are looking for

Strong practical experience in computer vision and perception engineering. Ideally experience across several of the following:

• Computer vision, machine learning or deep learning applied to real-world systems
• Object detection, multi-object tracking, re-identification or video understanding
• Experience working with live video, sensor data or deployed perception systems
• Strong Python and practical ML engineering experience
• Experience with PyTorch or similar deep learning frameworks
• Model optimisation, deployment or inference on edge hardware
• Comfortable working across research, engineering and product constraints
• Able to take ownership of ambiguous technical problems
• Strong communication skills and a low-ego engineering approach

Additional experience (Helpful, not essential):

• Robotics, autonomous systems or embodied AI
• Multi-camera or multi-sensor perception
• Vision-language models or open-vocabulary perception
• C++, CUDA, TensorRT, ONNX or similar deployment tooling
• Experience working in early-stage or fast-moving technical teams
• Exposure to outdoor, industrial or unstructured environments

Why this is worth considering

This is a chance to join an ambitious robotics AI company at an early stage, where the perception work will have a direct impact on the product and technical direction.

You would be working on systems that need to be fast, reliable and deployable, not just impressive in a demo.

The team is technical, ambitious and building for real-world use cases, with a strong focus on solving difficult engineering problems rather than adding unnecessary process.

For someone who wants ownership, technical depth and the chance to build perception systems that actually leave the lab, this should be a genuinely interesting opportunity.

Interested?

If this sounds relevant, apply here or message me directly.

Happy to share more context on the company, team and role once there is mutual interest.