Applied AI Scientist - On Site

Role overview

Autobrains Technologies is looking for a practical Applied AI Scientist to join its central R&D group in Munich. The role focuses on building the next generation of AI for autonomous driving, with work spanning both research and real-world deployment. You will help advance a multi-layer autonomy stack, with special emphasis on real-time predictive systems that support driving decisions.

This position suits someone who enjoys taking advanced ideas from papers and turning them into robust systems that can operate under strict performance and safety constraints.

Key responsibilities

• Lead the full lifecycle of driving-model development, starting with research review and early experimentation and continuing through production rollout.
• Build, test, and refine real-time predictive models, including vision-language-action models.
• Contribute to reasoning capabilities, especially VLA systems that support planning over different time horizons.
• Connect large-scale cloud training with embedded deployment by improving compression, quantization, speculative decoding, and fast inference for automotive edge hardware.
• Assess new techniques from the wider AI community and bring the most useful ones into the autonomy platform.
• Work closely with internal R&D colleagues to remove technical blockers, speed up delivery, and strengthen the overall engineering and research quality.

Requirements

• A Ph.D. in Computer Science, Electrical Engineering, Machine Learning, Robotics, or a closely related discipline; an exceptional M.Sc. profile may also be taken into account.
• A strong record of either publishing or deploying work in areas such as deep learning, computer vision, generative AI, reinforcement learning, or motion prediction.
• Clear evidence that you can translate research ideas into functioning software and not just theoretical concepts.
• Excellent Python programming ability, with C++ experience considered an advantage.
• Working knowledge of modern ML tooling and infrastructure, including PyTorch, ONNX, Triton, Dynamo, distributed training, and model optimization.
• Good grounding in probability, optimization, and statistics.
• Experience with CUDA or other low-level GPU performance tuning.
• Practical exposure to quantization, distillation, or efficient edge inference.
• Background in real-time, safety-sensitive, or embodied AI domains such as robotics, autonomous vehicles, or drones.
• Familiarity with foundation models across language, vision, tabular, or VLA use cases, especially when deployed on device.
• Knowledge of driving datasets, simulation setups, or sensor fusion pipelines is a plus.

Additional information

This is a hands-on technical role within core R&D, aimed at someone who can balance applied research with production engineering. The work centers on autonomous driving systems and may involve model development, optimization, and deployment under real-time constraints.