masonbrothers/openhouse-ai

Openhouse AI Challenge

This application takes a request made by a front end and stores it in a database.

Calls can then be made to the application to get data from the database.

To test

Please install Python 3 and add it to your terminal path.

After installing, run:

pip install -r requirements.txt

To run server, run:

./run-server.sh

After the server is running, to run tests, run:

python tests.py

REST API

POST /add-log

Posting to /add-log adds the log to the database. It returns a JSON string object with a message.

The format of the request must be:

{
    "userId": <string>,
    "sessionId": <string>,
    "actions": [
        ...,
        {
            "time": <iso-datetime as a string>,
            "type": <string>,
            "properties": <json object>
        },
        ...
    ]
}

GET /get-logs?userId=&startTime=&endTime=&type

Making a get request to the above URL will get the logs with the desired userId and type between the startTime and the endTime.

How to make application scalable

There are a few ways I could make this application scalable:

1. Use a different database. Sqlite is good for testing, but it only allows for one write at a time. Currently, this is a bottleneck as Sqlite will block on concurrent writes. Other SQL servers, like MSSQL, PostgreSQL or MySQL support multi-threaded writes.
2. Make Flask support asynchronous requests. Python is synchronous by default. This is can be done by using celery or some other multi-threaded queueing methods.
3. Launch multiple instances of the application. This could be done with AWS lambda or with multiple machines running multiple instances of the application. This would all be put behind a load balancer to route incoming requests to various machines.