Are you doing the firmware as well? Are you using pre-made hardware or developing custom hardware? The first place to start is at the hardware, because everything starts there and it may not support stuff you are thinking about right now. OTA is a very vast topic and can be easy in some cases if you are using existing platforms or can be quite an undertaking if you roll your own.
There are pre-built platforms like Particle where you get OTA and a basic platform for a nominal fee. They have pros and cons as any platform does. Plenty others as well.
MQTT is nice, but not all embedded platforms will support it for a variety of reasons. Security is a major component to embedded work as well.
My opinion having done these projects. Define the needs of the device and the features it needs to support now and over the next 3-5 years (best the business can predict). Then define the hardware based on those needs, that will let you know what software architecture(s) are reasonable. That will also let you look at various IoT platforms that exist like AWS etc and match them to your requirements. This up front process is way more important than it is with something like web development.
I can share war stories about what happens when you choose hardware via the MVP methodology like you would for a web project. Let's just say it never ends pretty, hardware is harder, takes more time and requires a more methodical approach then rapid iteration of typical software engineering. Of course, if you are just trying to check the temperature and humidity at a location and log them, it isn't very difficult and is solvable using off the shelf components which are low risk/low overhead.
Happy to answer questions if I can.