I encountered a similar issue writing CSV data from an application using Azure SDK. I observed that the Storage Events trigger wasn't firing when the application wrote to ADLSG2, but when I manually uploaded the file, the trigger fired.
I compared the properties of the files written by the application and the file I manually uploaded to ADLS and discovered their CONTENT-TYPE properties differed. The file created by the application was an application/octet-stream, while the file I uploaded was a text/csv. Since the application's data was written as a stream, it starts with "0 bytes and file size grows until all the data is written permanently"; consequently, if you created the trigger to "Ignore empty blobs", the file is actually empty at the first instance, and so ignored.
I solved this issue by setting "Ignore empty blobs" to No when creating the trigger.
Hope this helps!