AI Engineer: Multi-Model Agent Systems & Document AI

Xebiacee; Polska; Remote; PLN 180,000 - 360,000; Full time

Job summary

Xebia is seeking a senior AI/agent developer to design and implement multi-role decision patterns and model gateways for credit, KYC, and document processing use cases across Europe. You will build interfaces that securely route requests between models, support self-hosted deployments where required, and contribute to an evaluation framework for AI behavior.

You will collaborate with diverse engineering streams to ensure observability, governance, and auditable tool usage.

Qualifications

• Hands-on experience building AI or agent-based solutions.
• Experience designing or implementing LLM/AI agent workflows.
• Experience with model integration and routing, including consideration of cost, latency and risk.
• Experience designing controlled tool interfaces for AI agents.
• Knowledge of MCP-based tool interfaces or comparable permissioned agent-tool integration approaches.
• Practical experience with Document AI capabilities such as OCR, structured extraction, PII detection or masking.
• Experience with retrieval architectures using embeddings and vector or lexical search.
• Experience designing evaluation approaches for AI systems, including model or prompt evaluation.
• Understanding of observability requirements for agent behaviour.
• Ability to design AI solutions that operate through permissioned, logged and auditable interfaces.
• Ability to collaborate across prototype and production-oriented engineering streams.
• Ready to join within 2 weeks.

Responsibilities

• You will be building agents supporting credit, KYC/compliance, document handling and exception triage.
• Implementing multi-role decision patterns for high-impact cases, including analyst, critic and guardian roles.
• Developing a model gateway capable of routing requests between models according to cost, risk and latency.
• Supporting self-hosted models where data residency requirements make this necessary.
• Building MCP tool interfaces that restrict agent actions to permissioned and logged tools.
• Developing Document AI capabilities covering OCR, information extraction, PII detection and masking.
• Building retrieval and context capabilities using embeddings, vector search, lexical search and a knowledge graph.
• Developing an evaluation framework for AI behaviour, including evals, A/B testing of models and prompts, and observability.
• Working with the Foundation and Lab streams to ensure AI concepts can be implemented on the target platform.

Skills

AI/Agent development

LLM/AI workflows

Model integration and routing

Permissioned tool interfaces

MCP-based interfaces

Document AI (OCR, PII)

Embeddings / vector search

AI evaluation methods

Observability for AI

Auditable interfaces

Cross-team collaboration

Availability in two weeks

Job description

Xebia is seeking a senior AI/agent developer to design and implement multi-role decision patterns and model gateways for credit, KYC, and document processing use cases across Europe. You will build interfaces that securely route requests between models, support self-hosted deployments where required, and contribute to an evaluation framework for AI behavior.

You will collaborate with diverse engineering streams to ensure observability, governance, and auditable tool usage.