Computing Resources Management Platform for Tech Company

Background & Challenge

My client is a leading U.S.-based hardware manufacturer that relies extensively on AI across their entire business. As demand for AI computing resources keeps growing, existing infrastructure can no longer meet the needs of all teams efficiently.

To address this challenge, the company developed an internal platform that enables flexible allocation of GPU and compute resources based on team requirements and strategic business priorities.

Due to the confidential nature of the project, details cannot be disclosed.

My Role

Requirements Engineering and UX Design

I worked closely with the product team to turn fuzzy user needs into clear requirements. This involved breaking down complex technical and organizational topics and aligning stakeholders around a shared understanding of the product.

With low-fidelity wireframes and interactive prototypes in Figma, I made ideas tangible early in the process, enabling rapid feedback and user validation. In collaboration with a UI designer, we translated validated concepts into high-quality UI mockups using the client’s design system. We also supported the international development team throughout the implementation, ensuring that the designs were translated into a high-quality user experience.

Streamlining the Design Process with Claude Code

Alongside the “traditional” design work, I explored how AI could streamline the design and development workflow itself. I established a machine-readable context layer describing, user personas, and UX principles, and developed a custom Claude Code skill that transforms product requirements (PRDs) into initial interactive prototypes.

While the generated prototypes are not (yet) perfect, the benefits of this approach are already clear. Instead of relying on static click-through mockups, users and stakeholders can interact with functional prototypes integrated directly in the application. This reduces the time from PRD to a first prototype from days to minutes, yields more realistic feedback, and enables everyone on the team to quickly bring ideas to life and validate them.