Research Engineer (Robot Learning)

present of AI; München; Híbrido; EUR 70.000 - 100.000; Jornada completa

Descripción de la vacante

InLoop Robotics seeks a role focused on making warehouse robot policies faster and more reliable using data from robots operating at customer sites. The work centers on learning-based policies that handle picking and packing, with continuous improvement through real-world use and careful measurement.

A strong candidate has trained and fine-tuned robot foundation models with imitation or reinforcement learning, has put policies on real hardware, and collaborates with the deployment team to scale

Formación

• Trained and fine-tuned robot foundation models using imitation or reinforcement learning.
• Policies deployed on real hardware.
• Experience using data collected from robots operating at customer sites.

Responsabilidades

• Lead a research push to roll out 50+ robots and collaborate with the deployment team.
• Improve AI-driven policies for warehouse picking and packing using robot data.
• Benchmark and evaluate policy performance to ensure reliability at scale.
• Develop and maintain data pipelines for robot learning and evaluation.

Conocimientos

Imitation learning

Reinforcement learning

Foundation models fine-tuning

Robot policy evaluation

Human-in-the-loop

Descripción del empleo

The role centers on making warehouse robot policies faster and more reliable using data from robots already working in customer sites. A strong candidate has trained and fine-tuned robot foundation models with imitation or reinforcement learning, has put policies on real hardware, and is careful about how results get measured. Leading a research push to roll out 50 or more robots is part of the job, so working well with the deployment team matters too.

How it fits the AI era

The work is AI at its core: learned policies do the picking and packing, and the job is to improve them with every hour of real use. Tools for training, evaluation and data handling can speed up the loop, but judgment about what counts as a real improvement stays human. Deciding which failures matter to customers, and reading results with care, are not tasks to hand off.

How it builds AI experience

Someone in this role would gain hands-on experience running a full cycle of robot learning at growing scale, from collecting data on deployed robots through training to evaluation. The company runs vision-based policies with human teleoperation in the loop, so there is a chance to learn how human corrections feed back into better models. Building a similar background beforehand means training a policy on a real arm and writing up how it was tested.

Worth showing

• Policy running on real robot
• Imitation and RL fine-tuning
• Rigorous benchmarking and evaluation
• Data pipeline for robot learning
• Fast, robust inference at scale

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About InLoop Robotics

Robots-as-a-service automated picking and packing stations powered by continuous vision policies and human-in-the-loop teleoperation, per the company.