The need for microservices for ML
To understand the need for microservices-based architecture for ML applications, let's look at a hypothetical use case and go through various phases of developing a ML application for the use case.
Hypothetical use case
A large car repair facility needs a solution to estimate the number of cars in the facility and their accurate positions. A bunch of IP cameras is installed in the repair stations for monitoring the facility. Design an ML system to monitor and manage the car repair facility.
Stage 1 – Proof of concept (a monolith)
A quick PoC is developed in a typical case using available data points and applying ML to showcase and validate the use case and prove to the business stakeholders that ML can solve their problems or improve their business.
In our hypothetical use case, a monolith Python app is developed that does the following:
- Fetches streams from all cameras
- Determines the positions of cars (head...
