AI Engineer – AI Applications & Agents (all genders)

You turn AI prototypes into transparent applications that fit real business processes. The focus is on retrieval, tool integration and response quality. You work at the intersection of software development, data and security and help decide which tasks a model should take on and where deterministic logic is more reliable.

A knowledge assistant needs to find documents but may only return content that the respective person is authorised to access. You design ingestion, permission filters, retrieval and source display as a coherent flow. Agents require validated tool parameters and limited execution rights. Tests cover missing sources, manipulated documents and ambiguous questions. From the results, you determine whether chunking, reranking, prompts or the business process need to be adjusted.

• Develop RAG applications: prepare documents, select suitable chunking and retrieval methods and enforce permissions all the way through search.
• Implement LLM-supported workflows and agents; validate tool calls, limit execution rights and require human approval for critical actions.
• Build evaluations with realistic test cases and compare response quality, source attribution, hallucinations, runtime and costs.
• Integrate applications into existing systems through APIs; account for failure scenarios, fallbacks and observability from the outset.
• Develop permission-related retrieval tests and regression tests for source attribution and tool calls.
• Design timeouts, fallbacks and execution rights so that model errors cannot trigger uncontrolled actions.

What matters is demonstrable experience that is relevant to the role. It may come from appropriate vocational training, a degree, professional practice or a well-founded career change. We will align the specific level of responsibility with your knowledge and experience.

• Practical software development experience with Python or TypeScript and API-based applications.
• Understanding of retrieval, embeddings and the limitations of generative models; ability to assess quality using concrete tests.
• Careful handling of credentials, confidential documents and permissions.

This knowledge is helpful, but you do not need to have all of it at once. In your application, describe where you have already gained practical experience and which areas you would like to explore in greater depth.

• Experience with vector search, reranking or connecting internal company knowledge sources.
• Knowledge of prompt-injection defence and the evaluation of agent workflows.

A successful outcome is a transparently tested AI application with verifiable sources, limited actions and clear operating rules. You also explain to specialist stakeholders which requirements the solution deliberately does not fulfil with a language model.

This role can be based in Cologne or performed remotely. We will agree the specific form of collaboration, project requirements and any necessary customer appointments before you start.

This role is offered as permanent employment on a full-time or part-time basis. We will agree the responsibilities, working hours, start date and compensation to suit the role and your level of experience.

Show us in your application which experience or interests you bring to this role. A CV and relevant project examples, certificates or work samples will help us assess your professional background. Please do not send confidential customer data, credentials or protected source code.

Tell us your preferred working model, your possible start date and whether you would like to work full-time or part-time.

If you have questions about the role, contact our recruitment team at hr@otoko.com. Inclusion in the applicant pool is voluntary and takes place only with separate consent; it is not a prerequisite for reviewing your application.