Handling video uploads on a standard web tier is notoriously hazardous: buffering multi-megabyte payloads in server RAM exhausts heap memory and degrades request throughput. Watchly solves this constraint by piping inbound video uploads up to 100 MB directly into Cloudinary through Web ReadableStreams.
Built on Next.js 16 with the App Router, the application configures an elevated proxy body limit while delegating heavyweight transformations to Cloudinary using asynchronous eager processing (eager_async). Original payloads are compressed in the cloud, generating real-time metrics on byte reduction and bandwidth savings.
“Stream direct to the cloud broker — buffering 100 MB in server memory is an invitation to crash.”
Application state is anchored in PostgreSQL using Prisma ORM, storing video titles, public Cloudinary identifiers, original versus compressed byte counts, and playback durations. User authentication and protected route enforcement are handled via Clerk middleware, while optimized renditions and social share images deliver over CDN edges.
The investigation records that while upload streaming, database schemas, and transformation pipelines are implemented, asynchronous completion webhook handling and production deployment configs remain under active maintenance.