Selected work

Pipeline updates and performance reporting were split across 3 separate spreadsheets, making weekly reviews take days and ownership hard to trust.

Mapped sales stages, reporting rules, and operator roles before touching schema or UI.

Cut the weekly pipeline review cycle from ~2 days of spreadsheet reconciliation to ~15 minutes — single source of truth across sales, ops, and management.

Inbound orders, bookings, and chat updates were copied manually into spreadsheets across multiple branches, producing inconsistent records, delayed follow-up, and zero operational visibility for management.

Implemented web-order-to-sheet automation with validation and normalization.

Eliminated an estimated 60–80 hours/month of manual data entry across multi-branch operations. Booking + reporting now flow automatically into a single source of truth.

Users lacked a clear transition from AI chat prompts to meaningful search actions, and state sync issues reduced UX reliability.

Worked inside the existing WebView-based AI entry path so the feature fit the live app architecture.

Stabilized AI chat-to-search state sync in production — eliminated mismatch issues that broke the discovery flow for end users.

Separate repositories for related surfaces caused dependency drift and slowed coordinated releases.

Structured the repository with npm workspaces and clearer package boundaries.

Reduced release friction and dependency drift across the platform.

Social cards had uneven spacing and weak hierarchy, which made dense feed views harder to scan.

Reviewed spacing rhythm, card hierarchy, and interaction cues in the existing feed.

Improved feed scanability through low-risk UI fixes that fit the existing product language.