Replaced a stalled Firebase prototype with a type-safe tRPC and PostgreSQL platform for curriculum mapping, clinical hour tracking, and accreditation reporting.
Overview
A founder with preliminary technical knowledge retained me to take over and lead the complete development lifecycle for a competency-based education platform located at Milo. The application was designed to manage curriculum mapping, track clinical clock hours, and monitor individual and cohort learner progress against national academic accreditation standards. The project had stalled when the founder realized the underlying system requirements surpassed his initial implementation capabilities. The platform consisted of an unoptimized, basic Next.js setup sitting on top of an incomplete Firebase infrastructure. My task was to rapidly salvage the existing asset foundations, completely re-engineer the software architecture into a secure, multi-tenant enterprise application layer, and deliver robust role-based modules for faculty staff, students, and system administrators alongside a university billing infrastructure.
Architectural Strategy
The architectural strategy centered on replacing the fragile, low-overhead prototyping setup with a highly type-safe, relational database architecture designed to support strict data isolation and auditing. I migrated the primary data persistent layer away from Firebase, opting to deploy a relational PostgreSQL database managed by Drizzle ORM to enforce complex multi-tenant constraints smoothly. To streamline client-server communications and ensure absolute type safety across the entire application stack, I implemented a unified tRPC network layer coupled with React Query for efficient data caching, synchronization, and state management. The system required an intricate role-based access control architecture, which I designed from the ground up to prevent privilege escalation while allowing clear, separate system views and resource access boundaries for university administrators, teaching faculty, and active students.
Implementation Details
The technical execution began with a structural refactoring of the initial Next.js application, enforcing rigid runtime data verification patterns by implementing Zod schema validation across all data input boundaries. I engineered the complex curriculum mapping and learner tracking features from scratch, building logic that allowed students to log clinical hours, enabled faculty to review and approve supervised experiences, and allowed administrators to generate detailed compliance reports for academic accreditation evidence. To handle corporate educational billing, I integrated a multi-tier Stripe payment infrastructure, constructing automated webhook pipelines to handle subscription cycles, dynamic user seats, and university-level licensing configurations securely. Finally, I patched security holes present in the initial codebase, optimized relational query performance to prevent bottlenecks during cohort-wide logging rounds, and deployed the stabilized infrastructure to a production environment.
Outcome & Metrics
The architectural remediation successfully transformed an incomplete prototype into a production-ready enterprise platform on schedule, enabling Milo to confidently launch its commercial operations. By introducing a type-safe tRPC and relational PostgreSQL architecture, the platform eliminated the data synchronization glitches and security vulnerabilities that crippled the initial build. The system successfully managed the high-throughput demands of active student tracking, allowing thousands of simultaneous clinical records, experience approvals, and accreditation mapping evaluations to process seamlessly across multiple independent university systems without any performance degradation.