As a backend engineer, I designed and developed the whole backend architecture for this project. This project is used to monitor the employees' efficiency in a company/factory.

The project used a lot of technologies, especially in the backend. I used a kind of mini microservice architecture where I made Docker containers talk with each other through Kafka; in the real world, each container represents a real-life instance. The overall solution was designed to be self-hosted in the organization rather than the cloud, as certain factories will be in remote areas, and also, the cloud can get costlier over time.

The final product could blend in with any CCTV infrastructure. The project used ExpressJS to serve REST APIs to the client side, Kafka for server-to-server communication, Tensorflow to train our labelled objects, and OpenCV to get a live video feed from a CCTV camera. For each camera, we used to run a Docker container with the trained model, and that Docker container would detect the work the person is doing and also detect which person is doing it; all these model training can be done in the app itself.