AI Engineer - Insurance Domain - HYBRID

AI Engineer - Insurance Domain - HYBRID

Location:

Warren, New Jersey (US-NJ), United States (US)

• Design, fine-tune, and deploy Large Language Models (LLMs) for insurance-specific use cases including document intelligence, claims summarization, policy interpretation, and underwriting Q&A.
• Build Retrieval-Augmented Generation (RAG) pipelines using vector databases (e.g., Azure AI Search, Pinecone, Chroma
  
  DB) to ground LLM outputs in enterprise knowledge bases.
• Develop prompt engineering frameworks and systematic evaluation pipelines to ensure LLM output quality, consistency, and safety in regulated insurance contexts.
• Integrate LLM capabilities with internal data platforms via Lang Chain, Llama Index, or Semantic Kernel.
• Evaluate and benchmark foundational models (OpenAI GPT-4o, Azure OpenAI, Claude, Mistral, Llama) against insurance-specific tasks to guide platform selection.

• Architect and implement autonomous AI agents capable of multi-step reasoning, tool use, and decision-making for workflows such as FNOL triage, claims routing, policy lookup, and compliance checks.
• Build agentic frameworks using patterns such as ReAct, Chain-of-Thought, and Tool-Augmented Agents to handle complex, multi-turn insurance workflows.
• Design human-in-the-loop (HITL) checkpoints and escalation logic to ensure AI agents operate within defined risk and compliance boundaries.
• Integrate agents with internal APIs, data platforms, and enterprise systems using orchestration tools such as Azure Logic Apps, Apache Airflow, or Databricks Workflows.
• Develop guardrails, monitoring, and audit logging for all deployed agents to meet regulatory and governance standards.

• Build and maintain end-to-end MLOps pipelines covering model training, versioning, validation, deployment, and monitoring using MLflow, Azure ML, and Databricks.
• Implement CI/CD pipelines for ML models using Azure Dev Ops or Git Hub Actions, enabling reliable, repeatable model releases.
• Deploy models as REST APIs or batch inference services on Azure Kubernetes Service (AKS) or Azure Container Apps, ensuring scalability and low-latency response.
• Establish model monitoring frameworks to detect data drift, model degradation, and prediction anomalies in production.
• Manage the model registry and lineage tracking to maintain governance and auditability of all AI assets.
• Collaborate with data engineering teams to ensure feature pipelines are production-grade, versioned, and integrated with the Feature Store on Databricks or Azure ML.

• Work closely with business analysts, actuaries, underwriters, and claims professionals to translate domain requirements into AI solution designs.
• Participate in Agile/Scrum ceremonies including sprint planning, standups, and retrospectives as an active delivery contributor.
• Produce clear, well-structured technical documentation including solution designs, API specs, model cards, and deployment runbooks.
• Mentor junior engineers and contribute to internal AI engineering best practices and standards.

Bachelor's degree in Computer Science, Data Science, Machine Learning, Software Engineering, or a related quantitative field. Master’s degree is a plus.

• 3+ years of professional experience in AI/ML engineering, with demonstrated delivery of production-grade AI systems.
• 3+ years hands‑on experience building and deploying LLM‑powered applications using frameworks such as Lang Chain, Llama Index, or Semantic Kernel.
• 3+ years proven experience implementing MLOps pipelines in cloud environments (Azure preferred).
• 3+ years experience developing AI agents or automation workflows using agentic frameworks.
• 2+ years experience in financial services, insurance, or regulated industries is strongly preferred.

• Experience with P&C insurance workflows such as FNOL processing, claims triage, underwriting decisioning, or actuarial modeling.
• Familiarity with insurance regulatory requirements including NAIC guidelines and data privacy standards (CCPA, GDPR).
• Experience implementing responsible AI principles —…