AI Technical Program Manager

Position Summary
RK Industries is building an enterprise AI capability on Microsoft technologies to deliver measurable business value. Our governed Modern Data Platform (MDP) integrates ERP and operational data, with plans to mirror it into Microsoft Fabric to support AI and business solutions.
We’re hiring a Technical Program Manager to build and lead a team of AI engineers and agent developers, own RK’s technical AI roadmap, and remain hands-on in prototyping, solution design, and quality oversight.
Partnering with Strategic AI managers, business leaders, the Data team, and external partners, you’ll identify opportunities, translate business needs into AI solutions, ensure agents use governed MDP data, and transition early prototypes into production-ready solutions.
Primary Responsibilities
Build and Lead the Technical AI Team
Hire, onboard and develop a team of experienced AI engineers and agent builders.
Define roles, write job descriptions, run interview loops and set a high hiring bar.
Run the team day to day: set priorities and goals, assign work, run code and design reviews, manage performance and grow people's careers.
Define the team's operating model: how work moves from idea to prototype to production to support, and how prototypes hand off from partners or builders to engineers without rework.
Manage external AI development partners. Set acceptance criteria, review deliverables and reject or require rework of anything that doesn't meet standards.
Own the Technical AI Roadmap
Build and maintain RK's roadmap for technical AI work, sequenced by business value, data readiness and dependencies, in partnership with the Director and Strategic AI managers.
Lead technical delivery of AI data chat (Track 2), from the first pilot data agent on a well-modeled subject area to Teams-based adoption at scale. Support agent-enabled workflows in sandboxes and business solutions (Tracks 1 and 3). Choose the right tool for each case: Copilot Studio, Fabric data agents, Microsoft Foundry (prompt-based or custom-coded agents), Power Platform or traditional automation. Know when AI is not the answer.
Track the Microsoft AI and Fabric ecosystem (such as Foundry Agent Service, Fabric IQ and Microsoft 365 Copilot extensibility) and bring forward the capabilities that matter to RK.
Partner with the Business and Strategic AI
Work with Strategic AI managers and business stakeholders to find high-value agent opportunities, often with users who don't yet know what agents can do.
Lead process mapping and workflow discovery. Simplify or remove steps before adding AI, and decide which parts of a workflow an agent handles and which stay with people.
Turn business needs into clear requirements, success measures, baselines and acceptance criteria. Commit to business outcomes and report on them.
Communicate honestly with sponsors and leadership about progress, risks and agents that aren't performing as expected.
Build Enterprise-Grade Agents
Design and oversee agents in Microsoft Foundry and Copilot Studio, including tools and functions, enterprise grounding, integrations and Microsoft 365 / Teams deployment.
Design enterprise RAG solutions over documents such as construction specifications and drawings, covering authoritative sources, chunking, metadata, embeddings, hybrid and vector search, permissions and traceability.
Stay hands-on: build prototypes, lightweight coded agents, integrations and evaluation harnesses in Python, C# or SQL. Use AI coding assistants with discipline and validate everything they produce.
Use MDP and Fabric as the AI Data Foundation
Partner with the Data Architect and Data team so agents are grounded in governed MDP data mirrored into OneLake. No parallel pipelines from source systems.
Make sure agents use conformed cross-source objects so they don't double-count across ERPs. Raise data-model gaps, such as business states that aren't yet modeled, through the standard MDP intake process.
Help define and enforce row-level security and identity pass-through before any agent goes live, so an agent never shows a user data they aren't allowed to see.
Know the difference between analytical and operational questions, and design around data latency.
Quality, Evaluation and Responsible AI
Set the team's definition of production-ready: evaluation datasets, acceptance thresholds and measures for groundedness, accuracy, task adherence, safety and user experience.

Skills
Python, C#, SQL, Prototyping, ERP