Atlanta - Spring Boot and Tensorflow. Machine Learning as a Microservice
Speaker: DEV GUPTA SOFTWARE ENGINEER, PREDIKTO INC Many machine learning practitioners and data scientists use a common ML toolchain. This usually involves Python, Pandas, and some ML library. Recently a powerful and expressive tool has been added specially for state of the art neural network building, Tensorflow. Tensorflow allows data scientists to build deep and complex neural networks of all types (DNN, RNN's, LSTM, etc). However few data scientists have the skill sets to also build microservices for the enterprise with the necessary security, monitoring, and robustness required by many organizations. Luckily, there is a solution. Tensorflow allows the export of trained models and has API's for other languages, one of which is Java. In this talk, I will show some canonical Tensorflow code. I will then build a very simple microservice using Spring Boot and the Tensorflow API that allows for the respective model to be run on demand during an HTTP request.