Sentiment Analyzer is a ML model as a service made using Flask micro framework. The task is a binary classification problem where I employed CNNs to have a validation accuracy of 84.73% and validation loss of 0.3624. The application serves 2 pipeline: 1 pipeline where the model versions are trained using the jupyter notebook in the sentiment_analysis subfolder. The second pipeline is the inference pipeline where using a config file, Tensorflow Serving is used to deploy that model on a different docker container. The flask and tensorflow container are deployed on the same subnet and orchestrated via docker-compose.
Project link: https://github.com/kaumil/cnn_sentiment_analysis