Job Summary

This role engineers and ships production AI agents into the way a function already works. You embed inside functions across NetApp’s product organization, working inside their environment, systems, and tools. You determine what to build and how, the technical approach and the architecture, by working the problem alongside them rather than from a written specification. You build and ship the system end to end and own that it works and holds up in production - reliable and safe on the function’s real workflows - before rotating to the next function. The role sits within the Agentic Product Management team under the VP, Analytics Growth and AI Native Engineering.

The role combines three skillsets that usually live in three different people: the engineer who architects, writes, and ships the production system; the product-minded builder who can scope the right solution inside an unfamiliar function; and the consultant who can deliver inside a team that is not their own. The engineer is the spine - the person who writes the production code also embeds and scopes inside a function they have never worked in. Building AI agents is the baseline; what you deliver is a working system the host function runs and trusts in production.

A few principles shape how the team approaches its work:

• We start from outcomes, not from tasks. The first question on any engagement is what the process is for and what good looks like, not how to automate the current steps.
• We deliver working solutions, not recommendations. Every engagement ends with a prototype that real users can run.
• We design with audit trails. Every workflow is inspectable: which agent did what, against what input, with what reasoning.
• We expand trust deliberately. Agents start with bounded authority, and authority grows as evidence accumulates.

Key Responsibilities

• Embed inside a host function. Drop into a Product Group function such as product operations, program management, UX design, or product management - one that is not your own - for the duration of the engagement, and build the relationship with its leader and the operators who do the work today.
• Scope the technical solution. Working from the problem the host team surfaces, decide what to build and how - the approach, the architecture, where an agent removes real toil, and where human judgment, relationships, or accountability must stay in the loop.
• Build and ship the agent end to end. Put a working system into the host team’s live workflow using their real systems and data - prompts, tools, retrieval, sub-agents, and the integration code the deployment requires. The deliverable is production code running in the host team’s live workflow.
• Make quality measurable. Build for non-deterministic systems: evaluations, golden sets, and regression checks that make the agent’s quality defensible before it touches real work. You can show the system is correct from the evidence.
• Own that it works in production. Success is a system the host team runs and trusts on live work. You are accountable for the agent being correct and dependable under production conditions, and you run the engagement to that bar.
• Deliver like a consulting engagement. Scope the work, set expectations with host-team leadership, deliver on a timeline, and hand off with enablement so the team can operate the system after you leave.
• Rotate across teams and domains. Each engagement is time-boxed. You move across multiple host teams over the year, applying the same discipline to a new problem each time.
• Partner with security, IT, and data owners. In each host environment, work with the relevant owners to land agents responsibly under NetApp standards for access, privacy, and acceptable use, and to get the workflow approved.
• Contribute patterns back. Capture reusable components from each engagement - agent designs, prompt patterns, evaluation harnesses, integration approaches - and add them to the team’s shared infrastructure so later deployments move faster. Feed what you learn in the field back to the platform and roadmap, including missing capabilities and what should become paved-road infrastructure.

Education and Experience

• Typically requires 8+ years shipping production software, with a Bachelor’s degree in Computer Science, Engineering, or a related field, or equivalent practical experience.
• Hands-on fluency with current AI tooling, including LLMs, prompt and context engineering, tool and function calling, retrieval, agent frameworks, evaluation for non-deterministic systems, and protocols such as Model Context Protocol. Familiarity with Cursor and Claude Code is preferred, and experience with LangGraph and/or LangSmith is highly preferred; experience with other orchestration platforms and agent frameworks is welcome
• Strong problem discovery, structured thinking, and stakeholder management skills, including navigating security, IT, and data-owner review without losing the intent of the original design.
• Comfort operating in high ambiguity, including when the problem as described does not match the data and systems on the ground, and a preference for moving between engagements over owning one long-lived codebase. Breadth applied fast matters more here than deep specialization in one stack.
• Excellent written and verbal communication, including the ability to explain technical limits to non-technical stakeholders, disagree productively, and adjust as evidence evolves.