1. Create a pre-paid account mechanism as an account type or account option in Mifos (perhaps extending existing classes), zero fees.
2. Create a new API for consuming pre-paid account balances.  
3. Create a simple external model that mimics consumption by a household on a daily basis - imagined as a set of data received from solar panel (watts used, time of day, volts present).  
4. Load up pre-paid account by account transfer in P2P interface (my phone to your account) or via trusted agent - i.e. trusted agent receives cash and loads account on Mifos.

Providers of pay-as-you-go solar can use MIFOS for their operations. MFIs can become such providers.

At present MIfos X platform is shipped with Community-App (SPA) as default UI, but initial load time of the application high. And for users on narrow bandwidth internet it takes a considerable amount of time to load the app.

Instead of loading the app every time, we can convert the Community-App into a Chrome App and user can download them once and install it ion the machine, and it can used as thick client.

Community-App is completely developed on top of JavaScript, CSS and HTML. And Chrome apps are also written in same stack. Using grunt we should be able to repackage Community-App(SPA) into Community-App(Chrome App)

Users with slow internet connectivity can use this app (Installed on the machine) instead of loading Community-App (SPA) on the browser. Users will also have a continually updated app. This will also allow MFIs to potentially use Chromebooks as field-based devices for their branches.

The Mifos X platform as of today is a well structure but monolithic code base - one Git repo, one build, one WAR. This has worked out well for us, but as we grow, it could be interesting to introduce a proper SDK and plug-in concept, allowing contributors to build individual JARs that can be dropped into a certain directory of Mifos X to augment the core platform on well defined extension points and hooks. These modules would be loaded at run-time, not require rebuilding Mifos X, and would remain compatible when a customer upgrades their core platform version.

We currently see the SDK as something more relevant to the Java back-end, so in this project you would likely work less on our AngularJS UI (barring some configuration UI for the plug-ins probably); but if you do have any ideas how to make the front-end more modular and extensible as well, we would love to hear from you.

This project will have three axis/phases: 1. platform infrastructure framework kind of work, based on a very simple or fictitious example extension; 2. identifying the functionally which could require / be sensible to extend in Mifos X, extracting clear and documented Java APIs from it, and then make changes to existing functional platform services to make them extensible by plug-ins; 3. well document this - such plugins will only be writing if it's easy and clear how to do this. Note that the more technical part 1. should only take about 1/4 of your project, the focus will have to be putting the "theory" from 1. into "practice" in phases 2. & 3.