Case study

AmazonKepler Studio IDE — reframing the brief

Developer tools

Reframed a sidebar polish brief into a workflow-driven IDE redesign. The brief was 'fix icons.' The real problem: the sidebar exposed tools as a flat list while developers think in phases, and device registration blocked first launch for users who didn't need it. Shipped for Amazon's public September 2025 launch in 8 weeks.

Context

Kepler Studio is Amazon's VS Code extension for Fire TV developers. Adoption was around 25% and the team wanted 75% by end of 2025, tied to a public launch in September. I was brought in via Callstack as the contracted product designer across four streams.

Challenge

Sidebar was a flat list of tools — build/debug/performance under-discovered. Device registration blocked first launch for developers who had no Amazon integration. Marketplace icon used assets with licensing issues. Assets had to freeze by 8/15 for a 9/17 public launch.

My role

Lead product designer across four streams: sidebar IA, Build/Install/Run icons, KVD landing page, and marketplace extension icon. Direct with PM (Pavan Kumar Bhat), two engineering leads, and research partner (Page Diamond).

What I changed

  • Rebuilt sidebar around developer phases: Open → Build → Debug → Optimize. Build Mode collapsed into a compact dropdown defaulting to Debug, with active state surfaced in the status bar.
  • Made device registration optional. Primary CTA became 'Launch App.' Registration moved to secondary with a tooltip explaining when it's actually required (Amazon integrations like IAP). Added 'Skip & remind me later.'
  • Aligned icons to VS Code's native conventions after a licensing review with engineering — cleared the OSSM/Legal blocker before the freeze.
  • Cut the QR-based launch flow and complex onboarding to hit the 8/15 asset freeze. The registration reframe was the higher-leverage move.

Outcome

Shipped for the September 2025 public launch. Targets at handoff: −25% time to first tool, −60% registration mis-clicks, drop-off before first launch ≤12%. Post-launch measured data not available.