Lead AI Engineer - ML/LLM Ops & Agentic AI

Shift delivers AI agents that transform insurers' most critical work. By combining deep industry expertise and unmatched data resources, Shift provides proven results that have earned the trust of hundreds of the world's leading insurers. Our insurance-grade AI is accurate, explainable, and secure—empowering human experts to move with unmatched speed, total confidence, and a renewed focus on the people they serve. Your browser does not support the video tag. Our culture is built on innovation, trust, and a drive to transform the insurance industry through our SaaS platform. We come from more than 50 different countries and cultures and together we are creating the future of insurance. Learn more at www.shift-technology.com ABOUT THE ROLE At Shift, we are building the next generation of AI-driven platform capabilities. We are looking for an experienced, hands-on Lead AI Engineer in our Data & AI Platform Engineering team to drive our LLMOps roadmap and our most business-critical agentic use cases; someone who is deeply passionate about building, prototyping, and scaling modern AI/LLM systems, while leading a small team. This is a high-impact, high-visibility role for a senior builder who excels at the intersection of cutting-edge AI technologies (Agentic frameworks, LLM orchestration, MCP, GenAI, MLOps) and production-grade software engineering . You will lead a focused team of engineers and work directly with software engineers and product leaders to design, prototype, and ship state-of-the-art AI systems into production, while acting as the link with our Data Science team to align priorities with their needs and spread best practices. If you are a senior AI engineer looking for the next step in your career, where you can lead people and shape a roadmap without giving up hands-on coding and building , this role is for you. WHAT YOU'LL DO • Build agentic systems (hands-on): design and ship production-grade agentic workflows (LLM orchestration, tool calling, MCP, RAG, structured outputs), and define reference architectures that decide what a model handles and what deterministic code handles. • Own LLMOps standards: evaluation harnesses, prompt and agent versioning, tracing (OpenTelemetry), cost and token governance, and release gates, with tenant-aware patterns that respect client data isolation. • Evolve our ML platform: be the technical owner of our Databricks-based ML platform (Unity Catalog, MLflow, CI/CD to serving on AKS, monitoring), and extend it toward the agent and prompt lifecycle. • Run & Serve Open-Weight Models: Partner closely with our Cloud Ops team to design and deploy the architecture required to self-host open-weight foundation models internally. You will own the inference layer and model optimizations (latency, throughput), while collaborating seamlessly with Cloud Ops on GPU resource provisioning and cluster management. • Push the frontier: run time-boxed POCs on emerging techniques, such as relational foundation models for claims fraud detection or agentic ingestion of medical billing sources to generate claim edits, with clear success criteria and a path to production. • Lead and grow the team: coach and develop 2 to 5 people, raise the technical bar through design and code reviews, and partner with product, SMEs, architecture, security and SRE. WHAT WE ARE LOOKING FOR • Agentic AI expertise: proven experience building production agentic systems (Microsoft Agent Framework, OpenAI Agents SDK, PydanticAI, LangGraph or custom), and strong command of tool calling, MCP, RAG and evaluation. • MLOps and engineering: high proficiency in Python; hands-on experience with experiment tracking, model registry, serving and monitoring (ideally Databricks and MLflow); backend services on Azure, Kubernetes, Te