Strategic Analytics is a dynamic and growing team at Arch that drives innovation and transforms how the business operates. We build AI-first, agent-driven products that change how Arch underwrites, services, and learns from its book — combining frontier LLMs, multi-agent orchestration (MCP, A2A), retrieval-augmented generation, evaluation harnesses, and traditional ML. Our mission is broad: agentic automation, decision support, AI-driven insights, and the platform engineering required to make all of it production-grade.

We have a strong track record of success (productionalizing dozens of high quality Gen AI products over the last 3 years) - we aim to continue scaling these efforts and are seeking an AVP AI Engineering to lead the architecture & development of true multi-agent systems within Strategic Analytics. Reporting to the SVP of AI & Automation, you will design and operate orchestrations where agents communicate directly with one another — not just sequential, hand-off-driven workflows. After architecting the multiagent system, you will automate complex decisions by using Data Science frameworks/processes. This will happen in the system you create and production solutions must work at high levels of accuracy. You will partner with Implementation Engineering (IE) on the orchestration entry points and any infrastructure-side connectors.

Key Responsibilities

• Design our multi-agent orchestration patterns (master-orchestrator + specialized worker agents) using protocols such as MCP and A2A.

• This may be a blend of build & buy

• Lead end-to-end delivery of agentic underwriting and claims automations, from prototype through to production.

• Use precision/recall, calibration, confidence thresholds, error analysis, and business-impact measurement to determine when automation is safe to deploy

• Design decision frameworks that combine LLMs, retrieval, traditional ML, business rules, and human review.”

• Identify the conditions where the automation should be trusted, reviewed, or discarded

• Partner with IE on orchestration entry-point design (e.g., Azure Function endpoints, master-agent gateways) so the AE/IE seam is clean and scalable.

• Lead offshore engineers and team members on agentic patterns, prompt engineering, and reliability practices.

• Establish coding, evaluation, and observability standards for agentic systems within the AI & Automation Center of Excellence.

• Translate business intent into working agentic systems — not just systems that compile, but systems that deliver measurable business outcomes.

Required Skills / Experience

• 7+ years of software engineering and/or automation engineering experience.

• 3+ years of Data science experience

• 3+ years of people leadership experience

• Demonstrated experience building production grade multi-agent or agentic systems (beyond POC work).

• Strong track record of developing supervised learning models (ML and/or GLMs) that have a financially measurable impact on the business

• Strong experience sourcing & evaluating vendors

• Strong Python experience

• Resilient problem solving — comfortable with ambiguous problems and capable of breaking them into shippable increments.

• Strong written and verbal communication; able to operate across IE, AE, DS, and business stakeholders.

• Demonstrated, hands-on production experience (not POC-only) with the current AI stack: RAG, MCP / A2A or equivalent agent-to-agent protocols, agentic orchestration frameworks, and the ability to articulate where each is the right tool. Comfort scanning for and trialing new tooling as the space evolves.

Desired Skills / Experience

• P&C insurance domain familiarity — underwriting, claims, or submission lifecycle.

• Experience with retrieval-augmented generation (RAG), evaluation harnesses, and structured-output patterns.

• Cloud experience in Azure (preferred for our stack) and/or AWS; familiarity with private endpoints and enterprise-grade safeguards.

• Experience leading multidisciplinary teams (onshore + offshore) for technology delivery.

• Visible track record of self-directed learning in the AI space — side projects, contributions to agentic frameworks, write-ups, conference talks, or other evidence that the candidate is investing personal time staying ahead of the curve.

Education

• College degree in Computer Science, Software Engineering, Data Analytics, or equivalent practical experience.