What technologies to choose for a web application using neural networks?

Question

Python

Good afternoon!
Need advice on tech stack to help me. As part of the course work, you need to implement a machine learning algorithm. I decided to make a web application in python. I used to write neurons in C, I know Python has problems with performance. In general, I’m interested in what frameworks / libraries to use for the backend + machine learning (the algorithm is implemented from scratch, such a condition, so ready-made libs with neurons are not needed) and what will help me make it as productive as possible (and if possible, why exactly this, so as there is a lot of information, confused)

Answer 1

I have such a project. I used Tensorflow for ML + redis as an in-memory database and message queue (because ML is heavy, and it is better to take it out into a separate process) + Django.

Answer 2

Standard library stack for implementing ML algorithms from scratch in Python:
- Scipy,
- NumPy,
- Pandas,
- Scikit-learn.
For visualization - a bunch with D3.js.
More than enough for a course project.