4 months ago
Munich, GermanySenior
Responsibilities
- Design and build LLM-powered agentic systems from experimentation through production deployment.
- Develop retrieval and context pipelines, including RAG, hybrid search, and structured retrieval, over large technical and commercial datasets.
- Build evaluation, monitoring, and automated feedback systems to measure and improve AI performance.
- Create scalable pipelines for processing and structuring large volumes of unstructured documents and data.
- Own the architecture, reliability, scalability, and performance of production AI systems.
- Integrate AI capabilities deeply into Mercura’s product experience in collaboration with other engineers.
Requirements
- Proven experience architecting and building complex AI/LLM-powered systems or agents in production.
- Experience with model serving, agent-workflow orchestration, monitoring, and scaling AI workloads.
- Experience designing systems that process and structure large volumes of unstructured data.
- Experience developing evaluation and feedback loops for AI system performance.
- Python experience is required.
- Full-stack development experience with React, TypeScript, and FastAPI is a plus.
- Ability to take ownership, work with minimal guidance, iterate quickly, and solve problems end to end.
- Commitment to working in person in Munich in a high-intensity startup environment.
Benefits
- Real equity with favorable tax treatment.
- Free team dinners up to four times per week, UrbanSportsClub membership, a company bike, and team workouts three times per week.
- Easy relocation to Munich, with an apartment next to the WERK1 office available for around €800 per month.
- Company card and access to AI and productivity tools without approval requirements.
- High ownership, direct impact, founder-like autonomy, experienced colleagues, and mentorship from AI leaders.
- In-person work at the Mercura office in Munich.
Tech Stack
Categories
About Mercura
Mercura builds a SaaS CPQ platform for manufacturers, using product modeling, rules-based configuration, 3D visualization, and ERP/CRM integrations to turn complex technical requests into accurate quotes. It sells on a subscription basis to streamline sales-to-production handoffs and pricing. The privately held company is headquartered in Århus, Denmark, operates from Aarhus and Barcelona, and is a Y Combinator Winter 2025 alum planning expansion into the U.S.