Architecture

Edit or view the graph on Excalidraw using this link.

Aculei project is divided into three components: the backend, the frontend and the image classification module.

Cloudflare buckets

Cloudflare buckets are used to efficiently store images. The database only stores the URI of a given image. When backend is requested to return a set of images it retrieves metadata from the database and produces images’ URI . The frontend can retrieve images from cloudflare buckets using the URI provided by the backend.

Transformations

We chose to use Cloudflare’s transformations via url because the photos’ standard dimensions were far larger than needed. Essentially, transformations are altered copies of the same picture, such as downsampled, blurred, or resized. Transformations allow for a notable increase in archive load speed and take advantage of Cloudflare’s caching mechanism.

MongoDB

The database is hosted as free-tier MongoDB Atlas instance. This allow room for experiments and a lot of flexibily during development and testing.

Where is what?

• Frontend is deployed using Cloudflare pages to leverage all (amazing) Cloudflare functionalities.
• Backend is hosted on a Digital Ocean droplet running Ubuntu Linux. Backend code runs inside a docker container behind a Nginx Reverse Proxy.
• Database is hosted by MongoDB Atlas and accessed using Go drivers.