AI Agent Developer (LLM & Agent Systems)

NEDGEX is the corporate venture and investment unit of the NETZSCH Group, a family-owned German engineering company with over 150 years of history. It builds internal ventures and makes early-stage startup investments, providing capital along with hands-on strategic guidance and resources to help emerging ventures thrive. The team spans three locations in Germany and Brazil, combining engineering expertise with an entrepreneurial, innovation-driven culture.

Job Description

We are seeking a AI Agent Developer to advance the capabilities of our AI assistant, ideally with a background in natural sciences or mathematics. The core mission of this role is to extend and optimize our library of LLM tools and agent components. These form the integration layer between our LabV platform and state-of-the-art language models, providing structured access to laboratory data and workflows.

Qualifications

Responsibilities :

• Understand and analyze the laboratory use cases handled within LabV (e.g., quality analysis of recyclates).

• Design, extend, and refine LLM tools and agent workflows that support these use cases.

• Develop agent architectures (ReAct, Chain-of-Thought, Plan-Act, etc.) tailored to laboratory processes.

• Integrate LLMs and agent systems into our platform, ensuring robustness, correctness, and traceability.

• Optimize existing LLM-based functions to better support scientific and laboratory workflows.

• Collaborate closely with domain experts to translate laboratory requirements into effective agent behaviors.

• Prototype and validate agent behaviors directly against real laboratory data, refining them through hands-on iteration.

Must-haves :

• Strong experience with Python for AI/LLM development.

• Practical experience with modern agent frameworks and patterns (ReAct, CoT, Plan-Act, etc.).

• Hands-on experience with LLM/agent frameworks such as LangGraph, LangChain, or LlamaIndex.

• Experience with vector databases, embeddings, retrieval pipelines, and RAG architectures.

• Understanding of evaluation techniques for LLMs and agents.

• Knowledge of prompt engineering and document processing.

• Willingness to continuously learn and adapt to rapidly evolving LLM technologies.

Nice-to-haves:

• Experience with SQL, and with search engines such as OpenSearch or Elasticsearch.

• Exposure to laboratory, scientific, or industrial data domains.

Additional Information

CLT or PJ contract

Remote or hybrid work in Florianópolis