Rest API + Java. Scala, Hadoop + JavaScript Client : Deployment & Project Structure

Rest API + Java. Scala, Hadoop + JavaScript Client : Deployment & Project
Structure

Looking for inputs & best practice for new project :
Client side : JavaScript (probably AngularJS supporting tools) Server-side
& Rest API's : Java, Scala, Hadoop, Spring MVC 3 and other modules of
Spring too.
Github as repo
It will be Multi-module maven project. Normally maven web project as one
of the maven module + plus other maven sub-module, we put web resources
stuff i.e. .js (view+controller+model files), .css files, images and
develop REST endpoints (controller, action resources). It will be huge
project.
Which approach will be good one? 1. To put these JS(AngularJS) stuff into
here under web + Rest API using Spring MVC, deploy it or 2. Client side
has its own structure i.e. modular structure in sync with the server side,
maintain his own tree at Github and build, deploy cycle separate as client
side needs only REST APIs
The REST APIs can be used by internal teams or other components of the
project or may be by customers in future.
Build, deployment cycle separate for client side, separate for server-side
or together if we put client side code into web module of maven web
project. What are the best practices I can follow on the build life cycle
and deployment model. Deployment will be on Amazon +
CloudBees/CloudFoundry.