Android Platform

Plan

Android client app: Single Kotlin app shipped to the user.

• Data gathering: UI, user data collection and handling, and HTTPS client in Kotlin.

• ML: Call FedML's Java API from Kotlin for local training (Kotlin has first-class Java interoperability).

Server: Single modular Python server with a single database.

• Python is the language of choice to best support ML exploration.

• ML module:

  • Call FedML's Python API from the server for aggregation.

• Web module: gather and store data.

  • Django for HTTPS server and database interface (ORM).

  • PostgreSQL for the database.

HTTPS does not support broadcasting; we cannot assume the clients would always be on. So, I assume the clients will occasionally poll the server for new information. We only need to implement a REST API or something equivalent for communication.