ML Engineer, Agents & Reasoning

Skip to main contentEnglishEnglishBack to all jobsML Engineer, Agents & ReasoningBerlinFull-timePermanent employeeApply for this jobYour missionBuild agentic AI systems that reason, plan, and act inside real materials discovery workflows Most agent systems live in clean environments: browsers, codebases, or synthetic benchmarks. At Dunia, agents must reason about messy reality: experiments that fail, data that contradicts itself, and physical systems that don’t reset cleanly. As ML Engineer, Agents & Reasoning, you build the systems that make AI act responsibly inside that reality. You design agents that decide what to do next, use tools intelligently, recover from failure, and know when they don’t know. Your work sits at the boundary between cognition and control.  Your tasks will include: Build agentic decision-making systems for discovery Design and implement agentic systems that plan, reason, and act across materials discovery workflowsDevelop agents that operate over experiments, simulations, and scientific datasets, selecting next actions under uncertaintyDefine how autonomy is scoped, when humans stay in the loop, and how decisions are escalatedGround reasoning in scientific and physical reality Implement planning, control logic, and uncertainty-aware decision-making tailored to physical systemsEncode operational, experimental, and safety constraints directly into agent behaviorDefine stopping criteria, fallback strategies, and recovery mechanisms to prevent brittle behaviorTurn models into action Collaborate closely with AI researchers to embed predictive models into agent workflowsWork with lab, automation, and software teams to connect agents to real experimental and simulation systemsEnsure agent outputs translate into executable actions, not just recommendationsMeasure what matters Build evaluation frameworks that assess decision quality, learning efficiency, and system behavior, not just model accuracyAnalyze failure cases and iterate on system design based on real-world outcomesHelp define what “good decisions” mean in scientific discovery contextsShip reliable, production-grade systems Translate research concepts into robust, maintainable ML systemsInstrument agents with logging, monitoring, and diagnostics for observability and debuggingTake ownership of systems from prototype through deployment and operationYour profile4–8 years of experience building ML-driven or algorithmic decision-making systems in production or applied research settingsStrong background in scientific or structured data modeling, rather than language-first systemsExperience with planning, control, optimization, probabilistic reasoning, or decision-making under uncertaintyProficiency in modern ML frameworks (e.g.PyTorch, JAX) and strong general software engineering skillsComfortable owning systems end-to-end, from prototype to reliable operationAble to reason clearly about system behavior in complex, partially observable environmentsTechnically curious, with interest in physical systems, experiments, and real-world constraintsClear communicator who can work effectively across AI, engineering, and scientific teamsEnglish fluency;additional language desirableAbout usDunia, meaning “world” in over 20 languages, reflects our focus on building technologies that deliver abundance globally. By combining physics, AI, and automation, we accelerate materials discovery for next-generation energy and industrial systems. Our work helps make energy more accessible and materials more affordable and resilient while reshaping how science moves from idea to impact. Join us to work on problems where progress truly compounds.Need more convincing? -->  Watch this videoWe strive to create a diverse and inclusive workplace where everyone feels welcome and safe to be their authentic self. Non-traditional career paths are welcome and valued. If you share our vision, you can be certain that we want you to succeed. You might be just the right candidate for this or...