There is already codebase using GStreamer, there is no point creating a new one beside the Not invented here syndrome.
Wrting such application take too much time, I have been wrong in the past, but until you call main.c version 1.0, your project will never reach it. Backport your features in existing project, this is the best thing you can do.
Jason DeRose has been working on the dmedia library as a major component of Novacut, but it's only one part. It's what handles all the file magic and and edit data storage and sharing. The rendering backend and playback will all be built upon gstreamer. Novacut may seem like it's out of the blue and totally built from scratch, but there's actually a lot of smart code reuse architected in. Even dmedia lets desktop couch do most of the heavy lifting. GStreamer will let us build on all of the great work that's been put into making video decoding and encoding work so well on Linux, and when Novacut needs improvements there, any application building on GStreamer can benefit.
The problem, as I see it, is that making this cloud-aware is nearly the smallest part of the project. The big stumbling block to pretty much all of these projects is robustness with the NLE.
Certainly I'm glad you're reusing code, but is this a case of, as someone else said, NIH? Have you tried to speak with the PiTiVi/Openshot/Kdenlive/etc guys about the cloud intergration? I know the Pitivi guys are good ones to work with and are always asking for help.
Much better use of resources if the real goal is bettering the state of NLE in linux, but since this is a "company" venture I'm assuming at least part of the point is to create a product under your aegis.