Customer Support Intelligence API

A customer support intelligence API that interprets customer requests in the form of text and through the use of AI automate workflows predicts the category & confidence score of the request and generates a one-sentence summary.

Features

• Django + Django Ninja
• AI workflows using Bert classification & T5 summarization
• PostgreSQL database
• Celery + RabbitMQ Async workflow
• Gunicorn + Nginx Production setup
• Dockerization, Entrypoint & .env file
• Linting, Type Checking & Pytest

Installation & Setup

Clone repository

git clone https://github.com/JonathanB-96/Customer-Support-Intelligence.git
cd Customer-Support-Intelligence

Create virtual environment

python -m venv api
source api/bin/activate

Install dependencies

pip install -r requirements/base.txt

Run script

Important: Pre-trained models; bert-helpdesk-classifier & t5-helpdesk-summarize are not included in the repository, please request for them separately.

Local development

When setting local development for the first time update local.env file & run:

./run.sh build_local

If already built run:

./run.sh run_local

API should be up and running see: http://0.0.0.0:8000/api/docs for further access.

If adding or making changes to dependencies make sure to run:

./run.sh requirements

to update the requirements .txt files & re-build.

Production development

Similarly for production development update prd.env file while ensuring that the following field is set:

DJANGO_CSRF_TRUSTED_ORIGINS=http://localhost:1337

due to nginx app

& run:

./run.sh build_prd

Otherwise:

./run.sh run_prd

See http://localhost:1337/api/docs.

Running tests

For Linting, Type Checking & Pytest run:

./run.sh run_test

For further information on how to operate the run script see:

./run.sh help