About this role
Our vision is to transform how the world uses information to enrich life for all.
Join an inclusive team passionate about one thing: using their expertise in the relentless pursuit of innovation for customers and partners. The solutions we build help make everything from virtual reality experiences to breakthroughs in neural networks possible. We do it all while committing to integrity, sustainability, and giving back to our communities. Because doing so can fuel the very innovation we are pursuing.
Description
-
The Enterprise SSD Customer Validation Engineering (CVE) team currently develops validation scripts manually, with each new script typically taking 5–7 days end-to-end — covering feature scoping, characterization, test design and implementation, and verification. Before coding, an engineer must also understand the capabilities and limitations of different operating systems (Windows / Linux / FreeBSD) and vendor tools, which further lengthen the feature development process in CVE.
-
The intern will design and prototype an agentic AI framework, “CVE End-to-end Development Assistant”, that acts as a coding assistant for TMs. The intent is not to replace TMs but to let them focus on high-level test design while the AI handles low-level coding, repository navigation, and knowledge lookup. The solution follows a multi-agent architecture with an Orchestrator Agent that delegates subtasks to specialized sub-agents (Research Specialist, Test Designer, Developer, SSD Validation, and Challenger/Reviewer), and integrates with existing sources such as Jira, Confluence, JAMA, SharePoint and Bitbucket.
-
Project Scope
-
Learn about SSD validation workflows and the CVE repository structure, scripts and libraries.
-
Study modern agentic AI frameworks (e.g. Claude Agent SDK, Model Context Protocol, agent skills) and apply them to engineering automation.
-
Design a multi-agent architecture consisting of Orchestrator, Research Specialist, Test Designer, Developer, SSD Validation and Challenger agents.
-
Implement MCP connectors or agent skills to integrate with Jira, Confluence, TSM/JAMA, SharePoint, Bitbucket and local documents (PDF, DOCX, PPT).
-
Build a knowledge pool that captures OS-and tool-specific syntax, versioning and limitations across Windows, Linux and FreeBSD.
-
Enforce coding standards (Pylint, PEP8, CVE-specific rules) and automate version/copyright updates and standardized commit messages.
-
Benchmark the prototype against the current manual workflow to measure script development time reduction.
-
Troubleshooting, data analysis and iterative refinement of agent behaviors.
-
Document the design and perform knowledge transfer to the CVE team.
-
Deliverable
-
A working prototype of the “CVE End-to-end Development Assistant” multi-agent framework that reduces CVE script development cycle time from 5-7 days (feature scoping, test design, implementation and verification) to approximately