AI/ML Engineer (m/f/d)

What this role is about

Do you want to apply AI where data is incomplete and poorly documented? As an AI/ML Engineer at nodeline, you develop analysis and agent systems that turn this uncertainty into robust suggestions for data migration and transformation.

In long-running systems, it is often unclear what fields, tables, and relationships actually mean. Documentation is missing, naming is cryptic, and knowledge often exists in people’s heads rather than in specifications.

You approach the problem from two directions: text-based analysis of names and structures, and data-based analysis of distributions, cardinalities, formats, and referential patterns. On top of this, you build an agent system that brings these analyses together.

One principle is central: the system needs to make clear what it does not know. Confidence and uncertainty are part of the design so that humans can make informed decisions, especially in regulated environments.

Your responsibilities

Data Understanding: derive semantics from schemas, metadata, and the data itself

Schema Matching & Profiling: combine statistical patterns, embeddings, similarity methods, heuristics, and pre-trained models

Agent Systems: implement agents and agent orchestration

Confidence & Uncertainty: model uncertainty and support informed human decision-making

Evaluation: assess quality reliably even without complete ground truth

Production AI: integrate, test, and evolve models and analytical methods for production use

Real-World Data: work with anonymized, synthetic, or directly accessible data in customer environments when data is not allowed to leave the customer’s infrastructure

At the moment, we do not yet have our own training dataset, and the final analysis architecture has not been defined. In the first few months, the focus will therefore be on agent systems, statistical profiling, heuristics, embeddings, and zero-/few-shot approaches using pre-trained models. Custom ML models will become relevant once a sufficiently robust data foundation exists.

Your profile

Technical background:

• Degree in Computer Science, Business Informatics, Data Science, Mathematics, a comparable technical field, or equivalent practical experience
• Experience bringing models or ML components into production
• Strong Python skills at production level
• Experience working with little or no training data, e.g. zero-/few-shot approaches, embeddings, semantic similarity, statistical methods, and heuristics
• Experience evaluating systems without complete ground truth
• LLM integration beyond prompting: tool calling, structured outputs, handling non-determinism, and testing non-deterministic systems
• Solid understanding of databases and data modeling
• German and English at B2 level or higher, with confidence in both written and spoken communication.

How you work:

• Experimental and open-ended when exploring analytical approaches, but clean and robust when building production systems
• Independent, structured, and critical when assessing results
• Practical experience with coding agents or a serious interest in using them productively
• Documentation that makes knowledge transferable

Helpful, but not required: experience with schema matching, entity resolution, data profiling, or semantic type inference; data migrations or ETL; agent frameworks and their limitations; synthetic or anonymized data; or experience in industry, medical, or finance.

What you can expect

nodeline was founded in 2025. Until now, we have been a founder-led team, and we are currently expanding significantly through a funded project. You will join early and work with our CTO, engineering team, and pilot partners to determine which analytical approaches prove effective on real-world systems.

You will not be handed a clean, fully labeled dataset or a fixed target metric. Instead, you will help shape the methodology, architecture, and quality standards yourself.

• €62,000–€72,000 gross per year for a full-time role, depending on experience
• Remote or hybrid, with an office in Karlsruhe
• 30 days of paid vacation and flexible working hours
• Temporary work from other EU countries by agreement
• Time for your development: ½ day per week is reserved for professional learning and deepening your expertise; where workload allows, a full day
• Short decision paths and direct collaboration with the founders and engineering team
• Grow with the company: develop your area of responsibility based on your strengths and interests

We deliberately avoid siloed roles: everyone owns their area, understands the overall product, and supports the team where needed. Agents reduce workload, but they do not take over responsibility.

Interested?

No traditional cover letter required. Just tell us briefly what you have worked on most recently and what interests you about this role. A GitHub profile, LinkedIn profile, or CV is enough – apply via the platform or email us at [email protected].

If you recognize yourself in the role but do not meet every single requirement, we still encourage you to apply.

We welcome applications regardless of gender, age, ethnic background, disability, religion or belief, or sexual identity.

Privacy: Information on how we process your personal data as part of our recruitment process is available here.