Rest API for deploying ML models ⚡

A microservice built on Flask to deploy ML models as rest api.
This code can be used as a boilerplate for your next machine learning project deployment. Replace project/sentiment/ml_models/model.py with your custom model architecture file.

Steps to run

• Create virtual or conda environment and activate it Windows
  virtualenv restenv
restenv\Scripts\activate
  Linux
  python3 -m venv restenv
source restenv/bin/activate
• Install dependencies
  Edit line 15-17 in requirements.txt according to your OS
  pip install -r requirements.txt
• Install spacy model
  python -m spacy download en_core_web_sm
• To start the microservice in dev environment
  python manage.py start --env=dev
• To start the microservice in production environment
  python manage.py start
• To access swagger api, open http://localhost:5000/api/v1/
• To run tests
  python manage.py test
• Code coverage
  python manage.py cov

Input JSON and output JSON

Sample input json body

URL: http://localhost:5000/api/v1/sentiment

{
	"text": "i am so happy"
}

Sample output json body

{
    "status": "success",
    "message": {
        "text": "i am so happy",
        "sentiment": "positive",
        "confidence": 0.9986
    }
}

Input json body (bulk predict)

URL: http://localhost:5000/api/v1/sentiments

{
    "texts": [
        {
            "text": "i am so happy"
        },
        {
            "text": "I am so SAD"
        },
        {
            "text": "today is a Good day"
        }
    ]
}

Output json body

{
    "status": "success",
    "message": [
        {
            "status": "success",
            "text": "i am so happy",
            "sentiment": "positive",
            "confidence": 0.9986
        },
        {
            "status": "success",
            "text": "I am so SAD",
            "sentiment": "negative",
            "confidence": 0.9993
        },
        {
            "status": "success",
            "text": "today is a Good day",
            "sentiment": "positive",
            "confidence": 0.9989
        }
    ]
}

Swagger API

URL: http://localhost:5000/api/v1/

Sentiment Prediction using /sentiment endpoint

Work in progress...

TODOs ⏰

• add sentiment analysis model
• update api
• add tests
• add logging
• add load testing
• add database to store predictions
• dockerize the app
• add authentication

---