Agentic AI for Science - Developer (m/f/d)
datin GmbHHey there! We're datin GmbH, and we are building the "new power grid" for scientific innovation.
Traditional scientific knowledge is still largely hidden from LLMs because physical R&D data (like lab experiments, simulations, and equipment logs) is rarely recorded in a structured, machine-actionable way. Text alone isn’t rich enough to support automated discovery.
To bridge this gap, we have built an ontology-driven, schema-based knowledge graph management system. Now, we are taking it to the next level: building autonomous, goal-oriented AI agents that can interact directly with our graph databases, augment them with new data, and identify emerging patterns in physical science.
This role offers a unique opportunity to design production-grade AI agent systems from scratch, collaborating closely with experienced material scientists, tribologists, and software engineers. At datin, we value curiosity, impact, and trust, and we design our agent-driven workflows to empower scientists, not replace them.
Tasks
- Agentic Workflows: Design and build end-to-end agentic architectures. You will build tool-calling loops, memory layers, and execution environments that allow agents to query, update, and validate our graph databases.
- AI Infrastructure: Engineer, deploy, and maintain performant agent and LLM serving infrastructures both locally and in the cloud.
- Graph-Grounded LLMs: Fine-tune or optimize open-source LLMs to reliably translate natural language scientific requests into structured queries sent to our SDK and accurately traverse complex ontologies.
- Machine Learning for Science: Train and integrate specialized ML models to solve multi-objective optimization problems (e.g., predicting material properties or chemical reactions) that AI agents can use as tools.
- Semantic Digital Twins: Translate real-world physical workflows into semantically-typed knowledge graphs.
Requirements
- Technical Core: Deep practical experience with Agentic frameworks, orchestrators, or tool-use libraries.
- Software Engineering: Strong proficiency in Python and/or JavaScript, with a focus on writing clean, modular, and well-tested production code.
- Modeling Skills: Hands-on experience building, training, or fine-tuning models using machine learning frameworks like PyTorch or similar.
- Validation: Familiarity with SHACL, RDF, RDFS, OWL, and SPARQL or similar (like CYPHER) validation languages is a strong plus.
- Background: A degree in Computer Science, Information Science, or, Chemistry, Materials Science, Mechanical Engineering, or a related field.
- Mindset: You are meticulous and logical. You enjoy solving the
Opens the company's application page
Listed via
Arbeitnow
arbeitnow.com
Similar roles
Design & Tech
Related reads from TCHNX
Why AI Design Tools Are Quietly Replacing Junior Designers and What Actually Comes Next
AI tools promise efficiency, but London studios are discovering an unexpected paradox: automation creates new bottlenecks requiring precisely the expertise being eliminated. We investigate what's actually happening to entry-level design work.
The Inference Economy: Why AI’s Biggest Cost Shift Is Happening After Training
A major shift in AI economics is reshaping the industry. As training frontier models becomes more expensive and inference becomes dramatically cheaper, companies are being forced to rethink how they build, deploy, price, and monetise intelligent systems.
The Emergence of Small Language Models: Why Efficiency Is Overtaking Scale
As the AI industry confronts computational costs and environmental concerns, a new generation of compact models is proving that bigger isn't always better. Small language models are reshaping enterprise AI deployment.