AI Engineer — Workplace AI

Vanguard•Wayne, PA

2d•Hybrid

About The Position

Vanguard is establishing an enterprise Workplace AI capability to integrate agentic, generative, and assistive AI into the daily workflows of its crew members. This capability will be primarily built on the Microsoft 365 ecosystem, including Microsoft 365 Copilot, Copilot Studio, Azure AI Foundry, and Power Platform, with integration into Vanguard's existing data, identity, and security infrastructure. The role is for a hands-on AI Engineer responsible for implementing this capability, focusing on building and deploying agents and integrations using tools like Copilot Studio, Azure AI Foundry, Power Automate, and the Microsoft Graph. The engineer will also stay informed about advancements in other AI platforms such as AWS Bedrock, Amazon Q, Google Vertex AI, and Agentspace to incorporate best practices into the Microsoft 365 implementation. This role involves close collaboration with other team members on agent governance, environment strategy, program delivery, and platform alignment, and will be a key part of the emerging Workplace AI Fusion Team.

Requirements

8+ years of professional software / platform engineering experience, with 5+ years building production solutions on the Microsoft 365 and/or Azure stack.
Hands-on production experience with at least three of: Microsoft Copilot Studio, Azure AI Foundry / Azure OpenAI, Power Automate, Power Apps, Microsoft Graph, SharePoint Framework.
Demonstrable experience designing and shipping agentic or generative AI solutions — not just consuming a chat UI, but building, grounding, evaluating, and operating agents.
Working knowledge of at least one non-Microsoft AI platform (AWS Bedrock / Amazon Q, Google Vertex AI / Agentspace, OpenAI API direct, Anthropic API) sufficient to evaluate trade-offs.
Strong scripting and development skills in Python, PowerShell, and TypeScript/JavaScript; comfort with REST APIs, OAuth 2.0 / OIDC, and JSON schemas.
Experience implementing security and compliance controls in a regulated environment: Purview, Sentinel, Defender, Conditional Access, MIP, DLP.
Identity fluency with Entra ID (Azure AD) and federated IdPs (OKTA preferred).
Experience working in a financial services, healthcare, or other heavily regulated environment, or a credible plan to come up to speed quickly.
Bachelor's degree in Computer Science, Engineering, or equivalent practical experience.

Nice To Haves

Familiarity with Model Context Protocol (MCP), declarative agents, and the broader emerging agent-interop standards.
Experience with Confluence and Jira indexing / connector development for AI grounding.
Experience operating Power Platform Center of Excellence tooling and ALM patterns.
Comfort with infrastructure-as-code (Bicep, Terraform) and CI/CD for Power Platform / Foundry artifacts.
Background contributing to internal communities of practice, champions networks, or developer enablement programs.
Prior experience as a hands-on engineer in a Fusion Team (engineers + product + data + analysts) or AI Center of Excellence model.
Vendor certifications welcomed but not required: MS-102, AI-102, PL-400, PL-600, AZ-204, SC-100.

Responsibilities

Design, build, test, and deploy custom agents, plugins, and connectors in Microsoft Copilot Studio and Azure AI Foundry, integrated with M365 Copilot, Teams, Outlook, and SharePoint.
Build automation and orchestration in Power Automate and Power Platform, including custom connectors to Vanguard data sources (Confluence, Jira, ServiceNow, internal APIs).
Implement retrieval, grounding, and indexing patterns against SharePoint, Microsoft Graph, and approved third-party data sources.
Prototype and productionize agents that automate high-value crew workflows; iterate based on telemetry and user feedback.
Stay current on AWS Bedrock / Amazon Q / Strands / Agents for Bedrock and Google Vertex AI / Agentspace / ADK agent-building patterns, and translate applicable ideas into our Microsoft implementation.
Evaluate emerging frameworks (LangChain, LangGraph, Semantic Kernel, AutoGen, MCP) and recommend their fit within Vanguard's M365-first strategy.
Challenge the Microsoft default when a better pattern exists elsewhere, and defend the Microsoft choice when it is the right one.
Partner with the Workplace AI Engineer and Information Security to implement AI governance controls: data protection (Purview, MIP, Digital Guardian/DLP), model access controls, prompt and output logging, and agent lifecycle approvals.
Implement and operate Power Platform environment strategy (default, managed, dev/test/prod separation, DLP policies, capacity management).
Design IAM and least-privilege patterns for agent identities (Entra app registrations, managed identities, service principals, OKTA federation).
Build dashboards and telemetry for agent usage, latency, accuracy, cost, and policy violations using Log Analytics, Sentinel, Defender for Cloud Apps, and Power BI.
Define and operate the on-call model for production agents in partnership with Tier 3 Engineering.
Drive root-cause analysis for agent incidents and feed lessons back into platform standards.
Integrate Copilot and custom agents with Vanguard knowledge sources — Confluence, Jira, ServiceNow, SharePoint, Exchange, Teams, OneDrive — using approved connectors and Graph APIs.
Work with Identity engineering to align agent authentication with Entra ID and OKTA.
Coordinate with Data, Cloud, and Security domain teams on third-party data integration patterns.
Serve as Tier 3 escalation for complex Copilot Studio, Foundry, and Power Platform issues.
Provide deep technical consultation to Workplace Engineering and partner teams adopting agents.
Build reusable patterns, reference implementations, and internal playbooks that scale beyond a single engineer.
Lead proof-of-concept work for new Microsoft AI capabilities (Foundry agents, Copilot Tuning, MCP servers, declarative agents, agentic memory).
Contribute to the Workplace AI roadmap and operating model.
Partner with the Copilot Champions program and fireside-chat / train-the-trainer initiatives to translate engineering work into adoption outcomes.