From a serverless bottleneck to a modern, high-performance sports community platform
Challenge
The Challenge PLAI is a sports club management and community platform connecting members, clubs, and admins around news, events, and orders. As adoption grew — 50,000+ users, 2,000+ clubs, 6,000+ news articles, 57,000+ events, and 24,000+ orders — the original architecture began to buckle under its own success. Built on AWS Lambda + DynamoDB, the platform suffered from: High latencies on core read/write paths as data volume and concurrency grew Complex, hard-to-query relational data (clubs, groups, members, events, orders) forced into a NoSQL model, leading to data-modeling workarounds and duplicated writes Cold-start delays inherent to Lambda under bursty traffic (e.g., event RSVP windows, article publish spikes) Increasing engineering overhead to maintain and debug distributed serverless functions as feature complexity grew An aging UI that no longer matched the product's growing scope (comments, reactions, saved articles, rich event flows)
Solution
The Approach Rather than a patchwork fix, the team undertook a full re-architecture and redesign: Database: Migrated from DynamoDB to Amazon RDS, moving to a relational model that naturally fit PLAI's connected data (clubs ↔ groups ↔ members ↔ articles ↔ events ↔ orders), enabling proper joins, indexing, and query optimization instead of denormalized workarounds Application layer: Rebuilt on the MERN stack (MongoDB where document flexibility still made sense, Express, React, Node.js), replacing fragmented Lambda functions with a cohesive, maintainable service layer Infrastructure: Re-platformed on AWS, using managed compute and RDS instead of a pure FaaS model — trading cold-start unpredictability for consistent, tunable performance UI/UX: A ground-up redesign of the product experience — new interaction patterns for articles (save, react, comment, reply, mention), events, and club management, aligned to a modern design system
Outcome
The Outcome Reduced latency across core read/write paths Simplified data model → faster feature development velocity Improved reliability under concurrent load (event windows, article engagement spikes) Modernized UI driving higher engagement (saves, reactions, comments) Platform now supports continued growth across clubs, articles, events, and orders without architectural rework
90%
API/page latency improvement
85%
Reduction in support ticket
47%
Infrastructure cost reduction
50,000 clubs
Scale