Internship - Data Science & Robotics Machine Learning

Company Overview

RoboService focuses on solving the final deployment challenge in humanoid robotics. Hardware makers and base foundation models already exist, but those general-purpose intelligence systems still need to be adapted for the precise, repeatable demands of manufacturing. The company closes that gap by turning advanced models into reliable systems for production use.

Rather than creating foundation models from the ground up, RoboService works on putting existing state-of-the-art VLA models into practical use. The team refines them for better stability, speed, and accuracy in real manufacturing settings. Founded in July 2025 by repeat entrepreneurs with prior successful exits, the company is building software that helps robots move from prototypes and labs to factory floors.

Role Summary

The team is hiring an intern or working student to join its Robotics Machine Learning Engineering function. In this role, you will help connect manually controlled robotic systems with truly autonomous behavior. The work involves supporting robotics manufacturers that already provide solid hardware and a model framework, but need software help to adapt and tune models for a specific task.

Your main area of ownership will be building training workflows, starting with teleoperation data capture and continuing through VLA model fine-tuning and reinforcement learning for difficult manipulation tasks.

What You Will Do

• Develop pipelines that gather, organize, and inspect training data.
• Fine-tune VLA models, including both open-source systems and partner-provided proprietary models.
• Run and maintain data workflows, including teleoperation activities and large-scale dataset management for model training.
• Use reinforcement learning methods to improve robot actions and make them dependable in production-line environments.
• Work with advanced simulation tools such as MuJoCo, NVIDIA Isaac Sim, and Omniverse to train and validate agents before they are deployed to real hardware, following a sim-to-real approach.

Requirements

• You are currently enrolled in a Master’s or PhD program in Robotics, Computer Science, Data Science, Artificial Intelligence, or a closely related discipline.
• You have practical exposure to machine learning, deep learning, or embodied AI, along with hands-on robotics experience.
• You are comfortable programming in Python and working with deep learning libraries such as PyTorch, TensorFlow, or JAX.
• Experience with Transformers and VLA-style architectures is a strong advantage.
• You have worked with physics-based simulation tools, especially NVIDIA Isaac Sim or similar platforms such as Gazebo or MuJoCo.
• You can handle large datasets, automate data cleaning, and set up reliable training pipelines.
• You communicate well in English.

What We Offer

• Opportunity to transition into a full-time role after a successful internship.
• The role is based in Berlin and requires an in-person presence at the downtown office; relocation to Berlin is expected before the start date if you are not already there.
• Day-to-day access to advanced software and hardware technologies that are shaping the future of robotics.
• Work on real manufacturing projects with tangible impact.
• A fast-moving startup environment with ambitious growth plans.
• A highly skilled team where every contribution is valued.

Additional Information

This is a part-time remote role, though the company states that the team works from its Berlin office in person and expects candidates to move to Berlin before joining if they are not already there. No stipend or salary amount was specified. The posting does not mention the number of openings, duration, start date, or application deadline.