@KC it looks like you are having great challenges ahead :)!
To your main question, the direct answer is yes you can use Machine Learning in conjunction with an Azure IoT Central Solution which ingests data from an Azure IoT Edge device previously trained by your own Tensorflow Lite Models.
The bigger question is how do you want to do it and what are the current constraints you have + when do you want the actionable decisions coming from your connected device sensors to be made?
My advice is that you follow one of our Learning paths or Modules, for example:
- AI edge engineer
- Identify anomalies by routing data via IoT Hub to a built-in ML model in Azure Stream Analytics
You will learn how to use a trained model and clearly distinguish when you need an ML model running in the Edge vs in the Cloud.