iot VS mainflux

On the other hand, for maximum code reuse, you can use regular .NET 7, plus .NET IoT Libraries if you need sensors/access to GPIO and so on, plus Avalonia as a GUI, which is also using XAML, but incompatible with UWP.

Try the same program but with a hardware PWM library, e.g. https://pypi.org/project/rpi-hardware-pwm/

.NET nanoFramework has support for GPIO, I2C, SPI, PWM, ADC, DAC, Serial, 1-Wire. Also the API are aligned with .NET IoT making it easy for code reuse between development on a Raspberry Pi with .NET 6.0 and an MCU running .NET nanoFramework.

You can write machine learning application using C# with the help of https://dotnet.microsoft.com/en-us/apps/machinelearning-ai/ml-dotnet. Additionally IoT development as well using C# (https://github.com/dotnet/iot/tree/main/samples/force-sensitive-resistor)

Every sensor and module out there woks with on of the standard protocols like I2C or SPI and C# supports them. But everyone wants pre-made libraries/device binding and dealing with raw I2C and SPI is frustrating. .NET also has device binding libraries and most popular devices have binding there. Meadow team is also porting their binding to Linux.

For something more advanced you can look at this example but this is very advanced, its from an IoT platform written in Golang - https://github.com/mainflux/mainflux

There is lots of software to choose from, I suggest you have a look at Mainflux, and Mosquitto is popular for broker. Thingsboard is a turnkey solution, it's nice in some ways, but also very slow and heavy. Other options to research in this wikipedia article.