Head of AI Engineering (f/m/x)

Own and evolve our AI engineering function — transforming a 15–20 person ML team from research-heavy to a high-throughput, production-grade organization. You’ll partner with the CTO on strategy, build the platform that unifies LLM access, RAG, and backend services, and ship reliable, scalable AI features that change how banks work.
 
Key responsibilities 

• Team leadership and org build
  • Hire, mentor, and develop a high-performing team; set the technical bar, operating rhythms, and code/research review practices
  • Organize sub-teams (e.g., Core Modeling, AI Platform/Infra, Integrations) with clear ownership, SLOs, andon-call
  • Manage roadmap, capacity planning, and delivery across parallel initiatives
• Architecture and platform
  • Own the LLM gateway: unified APIs and proxy layers for multi-provider routing (OpenAI, Gemini, Bedrock), with rate limits, fallbacks, and cost tracking
  • Build high-performance RAG pipelines (ingestion, embeddings, vector stores, caching) with robust observability and safety guardrails
  • Partner with Java/NestJSteams to define clean async contracts, schemas, and eventing patterns; drive low-latency, scalable inference
• Model lifecycle and operations
  • Lead end-to-end model and prompt lifecycle: data curation, training/fine-tuning, evaluation, deployment, rollback
  • Establish LLMOps/MLOps: model/prompt registries, CI/CD, canary/A/B tests, offline/online evals, drift and cost monitoring
  • Optimizeinference throughput and cost (autoscaling, batching, quantization/distillation, caching)
• Strategy and collaboration
  • Translate company goals into an AI/ML roadmap with measurable outcomes; balance exploration with reliability and cost
  • Own build-vs-buy/vendor strategy for models, infrastructure, and data services; manage budgets and SLAs
• Governance and security
  • Implement data privacy, security, and compliance practices (RBAC, secrets, auditability); track prompt/model lineage and reproducibility
  • Define incident response, runbooks, and postmortems for AI features

• 5+ years as a backend engineer and 4+ years leading AI/ML engineering in production (10+ years total experience ideal)
• Deep architecture expertise in Java (JVM) and/or Node.js (NestJS), distributed systems, APIs, microservices, and messaging/streaming
• Hands-on with LLM stacks: orchestration (e.g.,LangChain/LlamaIndexor custom), vector DBs (Pinecone,Qdrant, FAISS), cloud AI (e.g., AWS Bedrock)
• Proven operation of systems at scale (millions of daily API calls) with strong SLOs, observability, and incident management
• MLOpsfoundations: model registries, experiment tracking, CI/CD, Kubernetes,IaC(e.g., Terraform), security best practices
• Excellent communication and stakeholder management; strong product sense focused on shipping user-facing feature 
• Fluent German and English for daily team collaboration, stakeholder management, and technical documentation

• Experience with GPU/accelerator serving and optimization (vLLM, TGI, Triton, ONNX Runtime)
• Cost optimization for LLM workloads (token budgets, dynamic routing, caching)
• Evaluation and safety/red-teaming for generative systems; startup/high-growth experience

• Platform: adoption of a unified LLM gateway; standardized observability and cost reporting
• Delivery: 2–3 user-facing AI features shipped with clear SLOs and measurable impact
• Reliability/cost: reduced average latency and cost per request; autoscaling and caching in place