Projects
MARKETPLACE ARCHITECTURE
Multi-Warehouse Fulfillment
ZALORA sellers with inventory across multiple locations had two options: cancel orders or relocate stock manually. Competitors supported up to 50 warehouse locations per seller. I led discovery across 12 seller groups, modeled the revenue case for OOS recovery and assortment expansion, and wrote the requirements for the product team.
€309K projected annual uplift
AI & AUTOMATION
AI Seller Support Automation
Indonesia's seller helpdesk was fully manual: limited hours, slow response times, and sellers waiting on basic answers. I developed the regional business case and led the Indonesia implementation of a third-party generative-AI chatbot, coordinating testing, rollout, implementation teams, and parent-company reporting. Once stabilized, it reached 68% ticket deflection, expanded coverage to 24/7 without additional headcount, and became the model for expansion to other markets.
68% ticket deflection rate
AI & AUTOMATION
Multi-Agent Sales & Onboarding
Sales and onboarding workflows were spread across calendars, CRM records, call transcripts, proposals, contracts, and client folders. I designed and built a three-agent n8n system connecting Calendly, Google Workspace, LLM APIs, and PandaDoc to automate lead follow-up, extract structured deal information from call transcripts, generate proposals and contracts, create client folders, and update onboarding status.
Three-agent end-to-end workflow
AI & AUTOMATION
AI Product Data Classification & Enrichment
Product teams relied on manual classification and attribute tagging across large SKU volumes, creating repetitive work and inconsistent outputs. I designed the classification logic, built a Python prototype, and developed the business case for an AI-assisted enrichment workflow. The approach included field-specific rules, confidence thresholds, structured outputs, and exception handling before engineering handoff.
70–80% projected manual-effort reduction
REVENUE ARCHITECTURE
Payment Handling Fee Design
Every major competitor charged a payment service fee. ZALORA Indonesia didn't, and the gap was recoverable. I built the business case: competitor benchmarking, three pricing scenarios across five payment methods, customer research, and checkout requirements for the product team; launched after handoff.
~€200K annual revenue generated
LOGISTICS & COST OPTIMIZATION
Cost-to-Serve Optimization
ZALORA's logistics costs looked manageable in aggregate. At the SKU and lane level, subsidy leakage was going undetected. I built a framework to find it, then designed the fixes: a flat-fee trucking model (58% subsidy reduction), multi-tier shipping fees across markets, and a cashless return chargeback system replacing manual deduction workflows.
€520K annual logistics cost savings
REVENUE RECOVERY
Return-to-Exchange Feature
When ZALORA customers wanted to swap a size or style, the only option was a refund. I led the launch of the Marketplace Exchange feature, which let customers convert returns into new orders without leaving the return flow. The initial rollout converted roughly 15% of returns into exchanges.
€695K annual revenue preserved
OPERATIONS & LOGISTICS
Routing Engine Optimization
At Paxel, activating a new delivery route took five days from identification to live operation, and the network's 130+ agent nodes needed a way to plan capacity as it grew. I wrote the business requirements for a routing optimization system that cut activation time to two days, and separately owned the capacity modeling approach and hub relocation strategy for the broader network, validating a payback period under one year.
60% reduction in route activation lead time
