Did you rsync in a loop? Or on demand?
I used Unison at Amazon, and it was the most seamless of all the options I tried. The two way sync was especially useful for being able to open up test results in a browser locally
You can write a script that uses the OS's native file watching system and triggers rsync. That's an approach I used and eventually everyone else on my team, not Amazon, also ended up using.
Wasn't very difficult to setup, on Linux at least.
I'd also recommend setting up some ignore rules for build artifacts, etc, which definitely increases complexity but avoids constantly syncing build artifacts you don't care about or with erroneous symlinks etc. IDK at that point I figured rsync was easier, as it's a more commonly used tool for other purposes, and I'd rather learn fewer things with huge functional overlap if given the choice.