AI Solutions Architect

LOCATION: Durham, NC (Hybrid) Victra is Verizon’s largest premium retailer in the United States. This role owns the technical architecture our agents are built on and the standards that keep them trustworthy in production. You will define the reference architecture, design the control plane (identity, scoped data access, audit, approval gates), set the evaluation and observability standard that separates production agents from demos, and provide the technical counterweight in every vendor conversation. Own the reference architecture for agentic AI on the Microsoft stack: Foundry project structure, Agent Service, retrieval and grounding patterns, and integration patterns to ServiceNow, point-of-sale, and HR systems Design and enforce the control plane: Entra ID and RBAC boundaries, scoped data access per agent, network isolation, audit trails, and approval gates, working with IT Security and the AI governance owners Define the evaluation and observability standard for production agents: pre-production evaluators and red teaming, continuous evaluation of sampled production traffic, drift detection, and alerting Make and document the platform routing call per use case: Copilot Studio as the governed low-code layer, Foundry as the pro-code runtime, or another platform when the use case genuinely belongs there, chosen by governance posture and outcome Own the framework and observability standardization decision: when to use Foundry-native capabilities (Agent Service, built-in evaluators, tracing) versus open-source components running on Azure (LangChain or LangGraph, Langfuse), and how the two coexist without fragmenting the platform or the audit trail Set the standards for AI-assisted development and coding agents used by the engineering team, including guardrails for what they may touch Provide technical review of all vendor and SOW work, including explicit lock-in and maintainability checks Produce per-initiative build-versus-buy recommendations against the function's standards Write reference implementations and prototype code; mentor early-career AI engineers on architecture and patterns Participate in incident review for production agents and feed lessons back into the architecture

Create an account to see the full posting, access our search engine, and more.