Agentic AI Engineer

Nasze wymagania: 3+ years of hands-on experience in generative AI, mainly text; audio/image experience is a plus. Proven experience with agentic frameworks – ideally POC agents built with AutoGen/AG2 (or similar frameworks). Experience hosting open-source LLMs locally; multi-GPU deployments are a strong plus. Solid background in unstructured document processing & retrieval; enterprise document experience is a plus. Strong Python skills. Experience with Git-based workflows, Docker and CI/CD (GitLab, Azure DevOps). Familiarity with Azure AI/ML services is a plus. Ability to work in English; Polish is a strong plus given the bilingual nature of our solutions. O projekcie: Design, build and operate autonomous AI agents and RAG systems in a regulated, enterprise environment (GDPR / EU AI Act, potentially medical devices). Own solutions end-to-end: from POC to production and decommissioning. Focus on building robust, compliant and observable agentic AI systems on top of local and cloud LLMs. Zakres obowiązków: Build and orchestrate autonomous AI agents using frameworks such as AutoGen/AG2, ADK or equivalents; familiarity with low-code tools (e.g. Langflow) is a plus. Implement and integrate agent tools (e.g. web search, MCP-based tools). Lead migration from external LLM APIs to locally hosted LLMs, with strong focus on performance and reliability. Design and maintain evaluation and monitoring for agent performance. Handle complex, heterogeneous document parsing (e.g. with Docling) and connect it to downstream agents. Implement on-prem RAG and advanced RAG pipelines, including visual document query and multimodal retrieval. Build solutions with bilingual support by design (Polish & English). Define and measure validation metrics, verify agents with target users, and iterate based on feedback. Use code generation tools (e.g. Codex, Tabby) to accelerate development. Architect, deploy and maintain end-to-end agentic LLM solutions on self-hosted and hybrid local/cloud environments. Own the full lifecycle: POC → production → maintenance → decommission. Use Git, GitLab, Azure DevOps and Docker for pipeline automation, CI/CD and reproducible builds. Implement and operate open-source and proprietary LLMs hosted in the cloud and on-prem. Instrument and monitor agents and models with Langfuse or equivalent observability tooling. Design solutions with compliance by design: GDPR, EU AI Act, and potential IVDR/MDR / FDA-equivalent requirements. Ensure robust secrets management, secure credential handling and environment configuration. Work in cross-functional teams (engineering, product, domain experts); Scrum/Kanban experience is a plus. Produce clear technical documentation; experience with IVDR/MDR / FDA-equivalent product documentation is an advantage. Define KPIs, run A/B tests and systematically validate model performance.