Physical AI Engineer

Your mission

Robots couldn’t handle soft materials. We fixed that.

Now we’re scaling it into reality, and we need someone who can make the impossible reliable.  

At sewts, we build the intelligence that lets robots manipulate deformable materials like textiles, towels, cables with human-like precision. Our systems are already deployed in production across Europe and expanding into the US.   We are building a foundational technology stack called sewts.JUPITER. JUPITER uses genAI to simulate realistically the behavior of deformable & soft objects to create corresponding computer vision models and grabbing strategies. Based on our physical AI, for the first time ever, robots are enabled to handle soft & deformable materials. With over €10M in funding we will bring automation into all untapped industries, like cloth manufacturing, textiles handling, recycling & e-commerce.   This is where AI leaves the screen and enters the physical world.

And in the physical world, nothing forgives you.

You make Physical AI work in reality.

Not on paper. Not in demos. Not in ideal conditions but with industrial required speed.   Gripping of deformable materials like textiles is a task most humans see as intuitive and simple.

However, this is a complicated task for robots as they lack the detailed and contextual understanding of where and with what level of sensitivity to handle these textiles.  

What You’ll Actually Do

• Build and scale AI systems that run on real robots in real environments;
• Develop deep learning models (Python, PyTorch) for computer vision and manipulation;
• Design and maintain end-to-end ML pipelines (training → evaluation → deployment);
• Own experiment tracking, reproducibility, and model lifecycle discipline;
• Push model performance toward robustness under real-world noise and edge cases;
• Automate training workflows, hyperparameter optimisation, and model iteration loops; 
• Work tightly with robotics,MLOps, and simulation teams;
• Help shape sewts.JUPITER into a production-grade Physical AI platform.

Your profile

Who You Are 

• 2-4 years' work experience in computer vision applications (object detection, classification); 
• Strong Python and PyTorch fundamentals;
• Experienced with ClearML, GCP, or similar ML infrastructure stacks;
• Deep understanding of model training, evaluation, and failure modes;
• Experience with ML pipelines, cloud infrastructure, or MLOps tooling;
• You care about what survives contact with reality, not just what looks good in experiments;
• You move fast, think clearly, and fix things when they break;
• You’re comfortable operating without perfect structure;
• You are fluent in English.

Nice to Have 

• Experience with robotics, industrial AI, or embodied systems;
• Vision Language Models or scene understanding exposure;
• Model optimization (TensorRT, quantization, PTQ);
• Experience in productionizing machine learning solutions.

Why us?

Why This Role 

• You build AI that physically acts in the world—not just predicts it;
• Your work runs directly on deployed robotic systems;
• Small team, extreme ownership, no abstraction layers;
• You shape a category: Physical AI for deformable materials;
• Real impact is visible in production, not dashboards;
• Munich-based, high-performance environment.

Setup & Benefits 

• Competitive salary + VSOP participation;
• Learning budget + fast growth environment;
• Deutschlandticket + EGYM Wellpass;
• Speciality coffee at the office;
• Flexible working setup;
• Regular team events, high trust, low bureaucracy.

Reality Check 

Things will break. Edge cases will win.  

If you want clean problems, this is not it. If you want to build intelligence that survives the real world— welcome to sewts. 

Your Talent Partner

We want to ensure a smooth recruitment process for you. If you have questions or would like to know more about this role, your Talent Partner, Fatima, is there to support you.       You may use the following channels to connect with her:       Email: fatima.sakhawat@sewts.de WhatsApp: +49 1577 74 44 352