Data Scientist / AI Engineer at Fuchs und Eule

Overview

Sole Data Scientist / AI Engineer across three production AI products at Fuchs und Eule, a Berlin greentech focused on residential energy efficiency and renovation. Two products were built from scratch and one was scaled from an early MVP to production.

LLM/RAG Search & Retrieval (built from scratch)

    Own the retrieval and evaluation stack for an LLM-first search product over 350k+ German/English technical documents (energy regulations, DIN standards, BAFA/KfW funding rules).

    Designed query routing, LLM-driven multi-step query decomposition, LangChain-based document chunking (markdown, recursive, semantic), BGE embeddings with cross-encoder reranking over Chroma, and k-tuning.

    Built a living query set and MLflow-tracked evaluation workflow covering retrieval precision and answer quality.

    Used daily by 40+ consultants and analysts.

Document AI for Regulatory Classification (built from scratch)

    Production ML pipeline that classifies invoice line items against complex external regulatory rules (BAFA/KfW funding eligibility).

    Auditable, structured outputs replacing manual expert review in a regulated environment.

Geospatial ML for Building Energy (scaled from early MVP to production)

    Automated building-energy pipeline using public 3D data (LoD2) to predict roof/wall/window components.

    Feeds physics-based DIN 18599 simulations via a kernel endpoint, producing estimates of energy state and financial KPIs (renovation cost, amortization, COâ‚‚ savings).

    Reduces energy-analyst effort from 3–8 hours to seconds.

Cross-functional & Product Ownership

    Owned the product layer across all three workstreams in the absence of a dedicated PM — driving roadmap, prioritization, stakeholder discovery, and acceptance criteria for ML deliverables.

    Presented technical roadmap and product capabilities to investors during a funding round.

    Collaborate daily with energy consultants, analysts, and software engineers to align technical solutions with practical needs.

  • Location: Berlin, Germany
  • Start Date: May 2025
  • End Date: Present
  • Relevant Technologies: Python, FastAPI, Pydantic, PyTest, Docker, AWS, PostgreSQL, LLM/RAG, LangChain, BGE embeddings, Chroma, MLflow, computational geometry, 3D modeling (LoD2), DIN 18599 simulation