Senior AI Forward Deployed Engineer (FDE) (m/f/d)

• It is not an internal IT or run role. You do not operate what you build - you hand it over.
• It is not a research position. We deploy frontier technology, we do not train frontier models. If your goal is to publish, this is the wrong team.
• It is not a data-science role. Notebooks and model metrics are not the deliverable; a system in production with users is.
• It is not a remote-first role. A significant part of the job is being physically where the problem is.
• It is not a graduate or junior entry point. We are hiring a small number of senior engineers who can own delivery alone in front of a demanding stakeholder.

We are looking for someone with...

• University degree in Computer Science, AI/ML, Mathematics, Engineering or a comparable quantitative field.
• 5+ years of hands-on AI/ML engineering, with a portfolio of systems that shipped and were used - not models that stayed in notebooks.
• Demonstrated customer-facing or stakeholder-facing delivery. You have been the engineer in the room, not behind a ticket queue.
• Deep expertise in the modern AI stack: Python, LLM APIs (OpenAI, Anthropic, Azure OpenAI, Gemini), RAG architectures, vector stores, embeddings, evaluation, and agent frameworks (LangGraph, Google ADK, Microsoft Agent Framework, OpenAI Agents SDK). PyTorch or TensorFlow where the problem needs it.
• Strong software engineering fundamentals: clean, tested code, CI/CD, containerisation (Docker, Kubernetes), API design, and a major cloud (Azure, AWS, GCP).
• Enough frontend capability to make a solution usable - you can build a decent interface without waiting for a designer.
• Experience with, or a serious appetite for, emerging paradigms: open-source agent runtimes, robotic and vision-language-action foundation models, multi-agent orchestration, tool-use agents.
• Willingness to work across enterprise platforms, not only cloud-native ones, and to learn platform-specific AI capabilities such as Agentforce and SAP Joule.
• Excellent communication: you can explain a technical trade-off to a non-technical stakeholder and build trust inside a short engagement.
• An entrepreneurial temperament: comfortable with ambiguity, energised by unproven technology, and able to deliver under time pressure without losing your judgement.
• Fluent English. German is an advantage for engagements based in Germany..

• Technology front runner. Be the first at Siemens to get emerging AI technology into production-grade systems. Validate what works. Document what does not, honestly and in public.
• Full-stack AI development. Design and build end to end: data ingestion and preparation, retrieval and model layer, agent orchestration, APIs, and the user-facing interface people actually touch. In this team, shipping a decent UI into an environment used to bad enterprise software is disproportionately powerful.
• Agentic AI engineering. Build agents and multi-agent systems on LangGraph, Google ADK, Microsoft Agent Framework, the OpenAI Agents SDK and Agentforce - with production reliability, evaluation and guardrails, not demo-quality behaviour.
• Cross-platform delivery. Work across SDC (Azure, AWS, GCP), and system-of-records, using the strengths of each rather than forcing everything into one.
• Embed with stakeholders. Sit with VPs, domain leads and process owners. Understand the process in detail - including the parts nobody documented - and turn that understanding into working software in days.
• Ship a kernel of value fast. Get something real and usable into the stakeholder's hands within the first two weeks. Credibility in this role is earned by visible velocity, not by a plan.
• Codify what works. Turn recurring patterns into reusable components, templates and building blocks that make the next engagement faster for the whole organisation.
• Handover excellence. Hand validated solutions to Domain teams to operate and to Agent Factory to industrialise, with clear documentation, architecture decision records, evaluation harnesses and operational runbooks.
• Field feedback that changes the platform. You are our eyes in the field. Feed back where models, frameworks and SDC primitives succeed and fail, with evidence - this is an expected output of the role, not a side effect.

Information Technology

Information Technology

Hybrid (Remote/Office)

• Amadora - Lisboa - Portugal

At Siemens, we build technology solutions to shape the world we live in. We transform industries and societies by combining the real and digital worlds. With over 300.000 of the world's most forward-thinking minds and the power of a presence in more than 190 countries, we make a truly global impact.

We take on challenges to make life easier, safer, and more sustainable for us and generations to come. We bring value to the business digitalization transition, from Portugal to the world, in areas such as Artificial Intelligence, Analytics & Business Intelligence, Application Lifecycle Management, Cybersecurity, IT Infrastructure Management, IT Project & Service Management, IT Strategy, User Experience, and many more.

With a decade of history and around 1.600 experts, the Portugal Tech Hub is the home of the new technologists - Dream Builders, Impact Creators & Future Makers.

First to deploy. First to solve. First to ship.

We are looking for Senior AI Engineers who are equal parts researcher and builder - the person who takes a new agent runtime, a frontier model or a foundation model that landed last month, and gets it doing real work inside a live Siemens environment before the industry has agreed whether it works. You build something usable in days, validate it with the people who will actually use it, harden it, and hand it over to run at scale.

Welcome to our world.

Forward Deployed Engineering is a specific operating model, not a job title we borrowed. It exists because Siemens now runs a genuinely powerful AI platform - the Siemens Data & AI Cloud (SDC) - and the people with the hardest problems worth solving with it sit in supply chain, finance, quality, sales and HR, not in engineering. FDEs close that gap by going to where the problem is.

An FDE is a customer-facing software engineer. You embed with a business team, you learn their process in uncomfortable detail, you build the thing that solves their actual problem, you get it into production, and then you hand it over and leave. You are not demoing, you are not advising, and you are not throwing a prototype over a wall.

Critically, you build on a platform. This matters more than it sounds. We are not a dev shop and we do not start from a blank page. Every solution is assembled from platform primitives that already exist, which is why a four-to-eight-week engagement is realistic and why what you build survives after you leave. If you find yourself rebuilding a primitive, that is a signal about the platform - and telling us so is part of the job.

• Siemens Data & AI Cloud (SDC) - our multi-cloud data and AI platform across Azure, AWS and GCP.
• Data layer - Snowflake and data products, Neo4j knowledge graphs, and the Knowledge Fabric.
• Agent layer - MCP & skill marketplace
• AI products - packaged AI capabilities to provide solutions and agents with the right tools
• Observability and governance - Arize for tracing, evaluation harnesses, guardrails and audit by default.

Your mission will be...

You are a hands-on technical leader working at the intersection of current AI research and enterprise-scale deployment. You work in four-to-eight-week engagements, embedded with business teams across Siemens, building AI systems against real, high-impact problems - intelligent automation in supply chain, decision support in finance, quality workflows on the factory floor, agentic processes across sales and HR.

You will be among the first engineers at Siemens to put emerging technology to work: open-source agent runtimes such as hermes, robotic and vision-language-action foundation models, NemoClaw and the NVIDIA stack, multi-agent orchestration and next-generation frontier models. Your job is to establish, with evidence, how they create value inside a complex industrial enterprise - and where they do not.

• Production adoption. Are real users using what you shipped, three months after you left?
• Measurable workflow impact. Cycle time, error rate, cost per transaction, throughput - a number the business owner agrees is theirs.
• Clean handover. Domain teams operate it without calling you
• Platform feedback that lands. Patterns you codified, primitives you pushed and evaluation evidence that changed what the platform team builds next.

• Small teams. Two to five people per engagement, high autonomy, minimal ceremony.
• Four to eight weeks, end to end. Discovery, scoping, architecture, build, production rollout, handover.
• Value inside the first two weeks. Something real and usable in the stakeholder's hands early, not a slide deck at the end.
• Onsite with the business. Roughly 40-50% of your time at the stakeholder's site - their factory, their office, their floor: context is the scarce resource and it does not travel over Teams.
• Then you hand over and leave. Operations go to the Domain team. You move to the next problem.