Backend​/Data Science Engineer

YUKKA turns unstructured news into measurable signal for financial institutions. We hold 20 years of quantified news data and deliver it through a cockpit, an API, reports and agents — to risk, investment, ESG and credit teams at banks and asset managers.

You'll work in our backend team, next to AI/NLP (Natural Language Processing) and frontend. You'll own services and pipelines end to end: APIs over our knowledge graph and document stores, data processing that turns raw news into insights, and the LLM-backed features going into the product. You will be reporting directly to the CTO.

• An API layer that serves graph, search and OLAP (Online Analytical Processing) queries reliably at production volume
• Pipelines that ingest, enrich and score millions of documents a day
• Agentic and LLM features — endpoints as a product surface
• Evaluation harnesses that tell us whether any of the above actually got better

• ~3–6 years building backend services in production
  . You write good Python and you know why your queries are slow.
• Comfort with data at volume: SQL and No
  
  SQL, streaming, batch processing, the operational reality of pipelines that break at 3am.
• AI fluency, in the practical sense: you've shipped something with an LLM behind it and know where it breaks. You can reason about retrieval quality, prompt and context design, cost and latency trade-offs, and how to evaluate output instead of eyeballing it. You use agentic coding tools as part of how you work, and you're sceptical enough to review what they produce.
• You can read an ML paper and decide whether it's relevant to us — you don't need to have trained the model yourself.
• Clear communication in English across a multi-disciplinary team.

Nice to have:
Kubernetes, graph databases, finance or NLP domain exposure, Rust.

A PhD, or a specific framework on your CV. We care that you code well, stay curious, and learn fast — we've hired that way deliberately.

• We value empathy, thoughtfulness and learning from our colleagues and our mistakes. Real innovation takes time, hard work and attention to detail, and we give ourselves room to think about hard problems. We are AI-first in practice, not slogan.
• Team language is English, German a plus.
• Offices in central Berlin, Hardenbergstrasse. Fixed salary plus possible stock options.