Skip to content

spiking_and_architecting

Jump to bottom
Stephen McCullough edited this page Mar 20, 2023 · 3 revisions

Spiking and Architecting

The app is currently at the drawing board stage. A few challenges exist and need flushed out before development can start. There is no point in creating tickets and all for this so this is just a collection of stuff at various stages of disarray.

I am sure this will get organised in time but at the moment it is just a mind dump.

App architecture

  • Rails 7
  • Postgres 12+
  • Redis
  • Sidekiq
  • No front end framework yet
  • Tailwind CSS to get more bang for my buck

A prototype app was put up that just has the sign in and sign out capabilities. I don't even think the user model was that sophisticated. I just wanted something up quickly. Decided to use [hatchbox.io] for deployment.

Image and video processing

I am thinking of an event driven app to do this. I want to be able to recognise faces and have some sort of intelligent model to try and extract meta information from a scanned in pic - such as who is in it (to a point and the decade). To do this I will probably use AWS Rekognition and S3 with a trigger to an upload lambda and then using boto3 to interface with the resources.

 +-------------+       +-------------+       +----------------+       +----------------+
 |             |       |             |       |                |       |                |
 |    S3       +------->  Lambda     +------->  Rekognition   +------->  State         |
 |  (Bucket)   |       | (Function)  |       |   (Service)    |       |   (Postgres?)  |
 |             |       |             |       |                |       |                |
 +------+------+       +------+------+       +----------------+       +----------------+
        |                     |
        |                     |
 +------+-------+   +---------+---------+
 | S3 Event      |  | Lambda Trigger    |
 | Notification  +--+ (S3:ObjectCreated)|
 +--------------+   +-------------------+
  • An image is uploaded to the S3 bucket.
  • The S3 bucket generates an event notification on the "S3:ObjectCreated" event.
  • The event notification triggers the Lambda function.
  • The Lambda function processes the image using the Amazon Rekognition service.
Clone this wiki locally