The final part of the PDF conversion project moves the heavy work into a serverless function. When a PDF reaches S3, Lambda receives the event and processes the document without keeping a dedicated conversion server online.

The function reads the uploaded file, turns each page into an image, and writes the generated files back to storage. A separate archive can then be assembled so the visitor has both individual page downloads and a single ZIP download.

An event-driven design is useful here because conversion demand is irregular. The application pays for processing when a file arrives, and the browser does not need to keep an open connection while the work happens.

The processing environment needs the right native libraries and enough temporary storage for the input and generated images. Packaging, memory allocation, timeouts, and file-size limits are therefore practical deployment concerns rather than details to postpone.

The browser can poll for generated images and the ZIP archive after the upload. This keeps the user experience simple: upload a document, show progress, reveal converted pages, and enable downloads when they are available.

Together with the React application and supporting components, this Lambda completes a small serverless product architecture. It demonstrates a repeatable pattern for direct uploads, asynchronous work, generated assets, and a lightweight front end.