PhD - Scalable and Efficient Reinforcement Learning Methods for Physical AI (f/m/div.)

Stellenbeschreibung:

Company Overview

At Bosch, we shape the future by inventing high‑quality technologies and services that spark enthusiasm and enrich people’s lives. Our promise to our associates is rock‑solid: we grow together, we enjoy our work, and we inspire each other.

Job Description

At Bosch Research, we are pioneering the next generation of Physical AI‑intelligent systems that can perceive, reason, and act robustly in the real world. From automated driving to robotics, our goal is to develop AI technologies that are not only highly capable but also safe, efficient, and scalable.

Despite remarkable progress in foundation models and end‑to‑end learning, today’s autonomous systems still struggle to generalize reliably to new situations and often require enormous amounts of data and computational resources. For Bosch products, however, AI solutions must deliver strong performance while operating under real‑world constraints such as limited compute, energy efficiency, safety, and reliability.

Explore new approaches for creating the next generation of efficient autonomous decision‑making systems.
Combine ideas from imitation learning, reinforcement learning, large‑scale simulation, world models, and model compression to develop AI agents (across several embodiments) that can continuously improve, adapt to novel situations, and transfer effectively from simulation to reality.
Enable resource‑efficient Physical AI by achieving the reliability and generalization of today's frontier AI systems while significantly reducing computational requirements for deployment on real devices.
Collaborate at the intersection of cutting‑edge AI research and real‑world industrial impact, working with leading experts from academia and industry to shape future intelligent products and services.

Qualifications

Education: excellent Master’s degree in Computer Science, Robotics, Mathematics, Physics or a comparable field.
Experience and Knowledge:
- Extensive hands‑on experience and theoretical understanding of reinforcement learning (RL), ideally with expertise in emerging behaviours, Safe RL, Offline RL, and/or Multi‑agent RL, combined with exceptional Python skills and deep knowledge of industry‑standard ML frameworks (e. g., PyTorch, Hugging Face, JAX, TensorFlow).
- Expertise in software development within larger teams and in deploying ML‑based systems for real‑world applications; background in MLOps and CI/CD is a plus.
- Good knowledge of C++ and related frameworks (e. g., ROS 2).
- Publications at top‑tier conferences in machine learning, robotics, or computer vision (e. g., NeurIPS, ICML, ICLR, CVPR, IROS, ICRA) are a plus.
Personality and Working Practice: methodical, structured, solution‑focused, quality‑mindful, proactive, collaborative, and driven to move beyond prototypes towards fully deployable solutions in close collaboration with business stakeholders.
Enthusiasm: deep enthusiasm for tackling complex technical challenges and developing novel, impactful solutions within interdisciplinary teams; passion for the synergistic potential between reinforcement learning and generative AI.
Languages: fluent in English; German is a plus.

Start Date

September 2026

EEO Statement

Diversity and inclusion are not just trends for us but are firmly anchored in our corporate culture. We welcome all applications, regardless of gender, age, disability, religion, ethnic origin or sexual identity.