Working Student

Description du poste:

Berlin, Germany AI Data Science LLM Python Software Development

KEA is hiring: Working Student AI / LLM (m/f/d) in Berlin
We're building an AI-powered data intelligence platform that turns natural language into verified database queries. A user types a question - our system resolves intent, retrieves the right variables from complex enterprise databases, generates and verifies SQL, and returns a precise answer. No SQL skills required. No analyst bottleneck.
The technology combines LLM orchestration, semantic retrieval, hybrid search, cross-encoder reranking, and self-correcting query generation across fully isolated multi-tenant environments. Our first clients are in market research, but the platform is domain-agnostic.
We're a small team in Berlin, close to launch, and looking for a working student who builds fast and thinks clearly.
What you'd work on
* Prompt engineering and LLM orchestration at the core of the product
* Retrieval logic - embeddings, BM25, reranking, making search precise
* Shipping features end-to-end that go live to real users
* Architecture decisions - you're in the room, not outside it
Stack
* Python, TypeScript, DuckDB, Supabase, n8n, Embeddings,
* BM25 + Semantic Hybrid Search, Cross-Encoder Reranking, RAG,
* NL-to-SQL, Intent Routing, Self-Correcting Generation,
* Multi-Tenant Isolation, Role-Based Access
What we look for
* Strong Python - the entire backend runs on it
* Interest in LLMs, retrieval, or making AI actually reliable
* Have built and shipped something (GitHub, side projects, prototypes)
* Learn fast, debug methodically, take ownership
* SQL is a plus - curiosity is non-negotiable
You don't need to check every box - but we expect you to be genuinely strong in at least one or two areas. One real project beats ten online courses.
Terms
* Berlin (on-site, not remote)
* 12-20h per week, flexible
* Up to EUR 25/h
* Start: ASAP
* Full-time down the road? Open to it - no pressure either way
How to apply
Short intro. One project you're proud of. Weekly availability.
Show us what you've built - that's worth more than any CV.
Email: johannes.nass@native-research.com
Full job page: https://ioatadp.github.io/kea-jobs