Machine Learning Engineer - Agents and Reasoning

About The Role

This position is a practical ML engineering opportunity at the cutting edge of agentic AI focused on scientific discovery. The selected candidate will develop systems that perform reasoning, planning, and autonomous actions within materials discovery processes, converting predictive models into dependable decision-making agents interfaced directly with physical experiments and laboratory automation. This role bridges AI research, software development, and experimental sciences, emphasizing autonomy, safety, and transparency throughout the entire discovery pipeline.

The employer is an early-stage deeptech startup dedicated to AI-led materials innovation and cleantech advancements, supported by experienced professionals and institutional investors. This role is located on-site in Berlin, Germany, and candidates must have legal working rights in Germany without requiring visa sponsorship.

Key Responsibilities

• Develop agentic systems capable of planning, reasoning, and acting within complex materials discovery workflows involving experimental data, simulations, and scientific datasets.
• Create decision-making frameworks that select appropriate next steps under uncertainty and intelligently balance autonomous actions with human oversight.
• Implement tailored planning, control mechanisms, and uncertainty-aware decision approaches to suit physical systems and experimental limitations.
• Integrate safety, operational, and experimental constraints directly into agent behavior, including defining stop conditions, fallback procedures, and recovery protocols to ensure robustness.
• Work collaboratively with AI researchers to embed predictive models into agents and translate model insights into executable instructions.
• Link agents with laboratory automation and software platforms ensuring decisions lead to tangible physical experiment outcomes.
• Implement extensive logging, monitoring, and diagnostic measures to increase system observability and assist troubleshooting.
• Design and maintain evaluation frameworks that assess decision quality and learning efficiency beyond traditional accuracy metrics.
• Investigate failure modes and refine system design based on operational experiences.
• Take end-to-end ownership of systems from prototyping through deployment and ongoing operation.

Candidate Profile and Requirements

• Between 4 and 8 years of hands-on experience in machine learning engineering, ideally working with autonomous agents or decision-making systems in real-world or research contexts.
• Proven expertise in building agent-based solutions that handle planning, uncertain action selection, and incorporate explicit stopping, fallback, and recovery logic.
• Strong background in delivering production-quality ML solutions with emphasis on observability via logging, monitoring, and diagnostic tools.
• Demonstrated experience integrating AI/ML models with laboratory automation, scientific hardware, or software control systems.
• Proficient in Python programming and familiar with major ML libraries such as PyTorch, TensorFlow, or JAX, along with comprehensive data processing libraries like NumPy and SciPy.
• Experience modeling scientific or structured data (as opposed to primarily natural language processing systems).
• Knowledge of implementing safety constraints and validation methodologies for autonomous systems operating in physical contexts.
• Excellent interpersonal and communication abilities to collaborate effectively across AI research, engineering, and laboratory groups.
• Fluency in English; additional language skills are advantageous.
• Legal eligibility to work in Germany without employer visa sponsorship is mandatory; visa sponsorship will not be offered.

Preferred Qualifications

• Familiarity with materials science, chemistry, or related physical science disciplines.
• Background in probabilistic methods such as Bayesian optimization or active learning.
• Knowledge of reinforcement learning, model-based planning, or control theory concepts.
• Additional European language proficiencies are beneficial.

Location and Terms

This is a full-time, on-site role based in Berlin, Germany requiring candidates to have a valid work permit for Germany without requiring visa sponsorship.