F&B E-Commerce Platform

An enterprise-grade online store system for a globally recognized F&B brand in Japan, covering the full commerce lifecycle: product management, inventory, ordering, discount & tax calculation, email notifications, and admin operations.

Key Contributions

• Reduced ElastiCache connections from 65,000+ to a maximum of 1,350 during peak traffic by applying the Singleton pattern and enforcing Lambda concurrency limits — eliminating connection exhaustion across the system.
• Improved email delivery success rate from 68% to 99.9% and resolved all EMFILE errors caused by exceeding Lambda’s file descriptor limit, by restructuring connection initialization to leverage Lambda’s invocation reuse mechanism.
• Enabled the mail system to handle up to 6,000 invocations at 60 concurrent executions, processing ~60,000 messages at an average of 500ms per invocation with zero errors.
• Designed and implemented a recursive BFS algorithm to correctly render nested gift-wrapped product structures on printed order receipts, handling arbitrary nesting depth and item aggregation rules.
• Solved a ¥1 rounding discrepancy in tax-inclusive discount calculations by isolating discount computation per tax category and applying a floor-based remainder allocation to the final item.
• Authored production-grade Node.js migration scripts for live data hotfixes, coordinating with the team to ensure accuracy and safety.

What I Did

• Developed downstream features: inventory management, product restocking, and transactional email notifications via SQS + SES + SNS.
• Developed upstream features: tax-inclusive pricing logic and multi-type discount calculation with accurate per-item allocation.
• Resolved critical Serverless infrastructure issues causing system-wide connection exhaustion and email delivery failures at scale.
• Optimized the order receipt printing system to correctly display deeply nested gift-wrapped product hierarchies using recursive BFS.
• Wrote a suite of unit tests covering all pricing and discount scenarios to safeguard calculation correctness during future maintenance.
• Maintained and extended PHP (Laravel) module within the platform.