Shadow Interviewer

The product already had a lot of depth. That was also the problem.

Technical interview products naturally involve dense workflows, different user roles, lots of states and information-heavy screens. As the platform grew, it needed a much stronger structure so users could focus on the interview itself instead of figuring out the interface.

The goal was not to simplify the product by removing capability. It was to make the complexity feel manageable.

This was not a redesign of a few key screens. The final scope crossed 200+ screens, which meant every decision had to work as a system.

I focused on repeatable patterns for navigation, forms, tables, panels, empty states, feedback and dense information. Instead of treating each screen as an isolated task, I spent more time defining the rules behind the screens.

A large part of Shadow Interviewer is used by technical users, so we intentionally borrowed interaction ideas from tools developers already understand.

IDE-inspired patterns helped organize dense interfaces, secondary panels, contextual information and actions without turning the product into a generic SaaS dashboard.

The goal was familiarity, not imitation: use mental models technical users already know, then adapt them to the interview workflow.

The platform also needed to support both light and dark themes. At this scale, that becomes a system problem rather than a simple visual requirement.

Every component, state, border, surface and hierarchy decision had to remain clear in both environments. Reusable patterns made the two themes feel like the same product instead of two separately designed versions.

With a product this large, assumptions compound quickly. We regularly tested and reviewed workflows, then adjusted layouts, hierarchy and interactions based on how people actually moved through the product.

Some of the most useful improvements were small: clearer actions, better grouping, more predictable navigation and less visual noise repeated across dozens of screens.

The biggest design problem wasn’t the number of screens. It was making 200+ screens feel like one product.

By the end of the revamp, Shadow Interviewer had a much more consistent product foundation that could support its complexity without exposing all of that complexity to the user.

• 200+ screens redesigned across a complex AI interview platform
• Simpler and more consistent workflows across different user needs
• Reusable UI patterns that made future product expansion easier
• IDE-inspired interactions that felt natural for technical users
• A coherent experience across both light and dark themes
• An interface refined through regular testing and iteration

Shadow Interviewer taught me a lot about designing large products where consistency matters as much as individual screen quality.

When a product reaches hundreds of screens, good design is less about making every page look interesting and more about building rules users can learn once and trust everywhere.

It pushed me to think more like a systems designer: simplify the workflow, reuse the right patterns and make complex software feel quieter than it actually is.