LivingTree

LivingTree is a school-family communication platform serving diverse communities where parents speak dozens of different languages. The platform's mission — connecting schools with families — was being undermined by a fragmented technical architecture. Multiple sub-sites (main app, teacher portal, admin dashboard, parent community) each had separate authentication systems requiring different credentials. The platform had no real-time translation capability, so non-English-speaking parents were effectively excluded from school communications. The SMS notification system could not handle throughput needed for school-wide announcements, and the provisioning system for 291 schools required extensive manual effort. Authentication-related support tickets were consuming significant resources.

We tackled LivingTree's challenges through three major initiatives. First, we implemented centralized OAuth SSO using OAuth 2.0 with PKCE flow across all sub-sites, supporting email/password and social login with automatic account linking. Second, we built a real-time translation pipeline supporting 15+ languages using AWS Translate with a caching layer that stores previously translated phrases to reduce API costs. The system detects parent language preferences and automatically translates all school communications. Third, we redesigned the SMS and short link engines using a queue-based architecture for scale, with trackable URLs providing engagement analytics. We also built a graph-based recommendation system using Neo4j for community features and streamlined the student provisioning system to automatically sync data across all 291 schools.

The platform serves 20,000+ active users with full multi-language support. OAuth SSO reduced authentication support tickets by 70% and increased daily active usage by 40%. Schools in diverse communities reported 55% higher engagement from non-English-speaking families. The translation caching layer handles 80% of requests without API calls. Graph-based Neo4j recommendations delivered results in 15ms compared to 3-5 seconds with PostgreSQL. The provisioning system reduced manual effort by 80% across 291 schools. The team maintained 98% code coverage while resolving an average of 25 bugs per sprint.