Amazon's inbound transportation network served 1M+ sellers shipping 20M+ units per day. Sellers were booking delivery appointments by picking arbitrary dates without visibility into carrier capacity or fulfillment center readiness. The result: missed deliveries, wasted carrier trips, stockouts, and a cascade of costs across the supply chain.
Operations teams were unilaterally rescheduling 8-10% of appointments to manage congestion, but only ~20% of those reschedules stuck. Carriers were already in transit. Sellers were blindsided. The system had no handshake between the parties who needed to agree.
I owned this end-to-end. I pulled six months of historical data, organized a three-day Kaizen with directors from operations, transportation, tech, and planning, and diagnosed the core issue: reschedules happened without knowing whether freight was in transit or what the shipment priority was.
I designed a capacity-matching transportation workflow that let sellers book against real-time carrier capacity. The UX showed green/red date indicators backed by actual FC capacity, with chargeback-free backup dates as safeguards. A new handshake workflow required confirmation from both seller and carrier before any reschedule took effect.
I built a self-service appointment scheduling portal (pre-filled details, real-time capacity view) and launched APIs (10K calls/day, <500ms P99 latency) enabling TMS integration for 3P sellers.
This wasn't just a feature launch. It was a fundamental re-architecture of how a two-sided marketplace handles logistics coordination under uncertainty. The workflow scaled from 5 to 50+ sellers processing 1M+ units/week, achieved 90%+ CSAT, and became the foundation for Amazon's inbound appointment platform going forward.
I also managed 2 PMs through this period, established intake and launch playbooks adopted across 5 product teams, and compressed roadmap alignment timelines from 4 weeks to 2.