Lead Agentic AI Architect

Verisk•Jersey City, NJ

2h•Hybrid

About The Position

We are seeking a hands-on technical leader to define the architecture for, and deliver, production agentic AI solutions across our software portfolio. You will build critical components, guide engineering teams and establish reusable capabilities for AI-powered products, business workflows and AI-enabled software development, partnering with Product, Data, QA, Security and Platform teams to deliver reliable solutions with measurable business value. Why this role: Shape how agentic AI is applied across insurance products and workflows

Requirements

Engineering leadership. 10+ years in software engineering, including 4+ years leading AI and agentic architecture or technical delivery across multiple teams, or equivalent demonstrated expertise.
Production AI delivery. Has deployed and operated multiple LLM or agentic application in production; strong Python, APIs, distributed systems, RAG and agent orchestration skills.
Cloud and collaboration. Experience with cloud deployment, containers, CI/CD, evaluation and observability; able to explain trade-offs and influence technical and business stakeholders.

Nice To Haves

AWS and Amazon Bedrock, including AgentCore
LangGraph, Stands or comparable agent frameworks
MCP
semantic layers, ontologies or knowledge graphs
insurance or other regulated data products
AI-enabled SDLC practices.

Responsibilities

Set portfolio architecture. Define reference architectures, reusable agent services and engineering standards. Select models, frameworks and build-versus-buy options based on quality, cost and business needs.
Lead through implementation. Write production code, review designs and pull requests, resolve complex technical issues and coach teams through delivery and adoption.
Design dependable workflows. Implement tool calling, orchestration, context and memory, retries, execution limits and human approvals. Choose simple automation, a single agent or multiple agents as the problem warrants.
Connect enterprise knowledge. Build Agentic, retrieval-augmented generation (RAG), structured data access and API or Model Context Protocol (MCP) integrations with permission-aware retrieval and source traceability.
Establish AI evaluation. Partner with QA and domain experts on test datasets, task-success measures, groundedness, tool-call accuracy, adversarial testing and release acceptance criteria.
Own production readiness. Implement CI/CD, versioning, monitoring, agent tracing, fallbacks and rollback. Track reliability, latency and cost per successful task, and support incident resolution.
Embed security and governance. Apply least-privilege tool access, sensitive-data protection, prompt-injection defenses, auditability and approval controls with Security and governance teams.
Advance AI-enabled development. Introduce and govern AI-assisted engineering practices such as coding agents, AI code review and test generation, and measure their effect on delivery speed and quality.
Drive portfolio outcomes. Prioritize use cases with Product, maintain the technical roadmap and share reusable components and lessons across teams to improve delivery speed and customer value.