FundsCorner

For a slide deck view of this content, visit https://www.slideshare.net/JeyabalajiSubramania/flutter-at-fundscornerpptx

Context: Recently we were looking for a tool to build cross-platform mobile apps. The primary goals for us were two fold:

1. Ability to roll out the mobile app fast. Being in the FinTech segment, our focus is more on usability & accuracy and less on the flashiness of the app in itself
2. Our web development team must be able to build mobile apps. The UI & UX fundamentals are pretty much the same.

With the above in mind, we evaluated React Native, Vue Nativescript and Flutter. While we were able to build fast in all these three choices, we chose Flutter for the following reasons:

Pre-built widgets: All the standard widgets that are required for us to build a functional app were readily available, & required minimal or no tweaking! It was pretty much like cooking up something on the web with Vue & Vuetify, which offer the fastest time frame from code to reality. The key differentiation Flutter offers over it's rivals is the native feel you get on all the widgets. No one can figure out whether it was built in Native android or Flutter.

Availability of Pre-built widgets in Flutter makes it a natural choice for going the fastest from design to reality.

Easy programming constructs & Hot Re-load: The component coding for Flutter is done through Dart. It is kind of a cross between Java & JavaScript. It is easy for the developers. I found asynchronous programming in Dart a breeze! Dart is one of the key reasons why you would build an app in record time with Flutter. Also, you will love the hot reload feature in Flutter, through which you can immediately validate the user interface and interactions.

Hot Re-load is one of the key features that make development in Flutter a breeze.

Rich set of plugins & great documentation: Flutter eco-system has matured over a period of time. We were able to easily find solutions to various problems & all the plugins worked without breaking anything. For example, we wanted to build a web view for integrating with a Payment link & flutter_webview_plugin was readily available and we were zooming in less than 30 minutes!

With great documentation and eco-system, you are always a plugin or a widget away from completing your functionality!

Great support for Reactive State Management: We were spoilt for choices when we looked at the various options for implementing Reactive statement management. After looking at a number of options, we settled with RxDart and Provider Consumer (Bloc) pattern to implement reactive statement management.

You will be able to apply your hard earned reactive state management skills in Flutter seamlessly & built beautiful reactive apps.

Easy integration with Android Native SDKs: Flutter provides a platform interface to integrate with native SDKs. Being in the FinTech industry, we were required to integrate with a number of industry-standard SDKS for payments & KYC, which were available only in Native. We were to connect with these SDKs and code with ease with the platform interface.

In the end, we were able to build and release an end-to-end, material design compliant and functionality rich Borrower app within a matter of one month and release it for Beta preview!

With the announcement of Flutter for Web in this year's google I/O, I think Flutter is going to go big and will shake up the world of cross-platform development.

At FundsCorner, when we set out to pick up the front-end tech stack (around Dec 2017), we drove our decision based on the following considerations:

(1) We were clear that we will NOT have a hybrid app. We will start with Responsive Web & once there is traction, we will rollout our Android App. However, we wanted to ensure that the users have a consistent experience on both the Web & the App. So, the front-end framework must also have a material design component library which we can choose from.

(2) Before joining FundsCorner as a CTO, I had already worked with Angular. I enjoyed working with Angular, but I felt that I must choose something that will provide us with the fastest time from Concept to Reality.

(3) I am strong proponent of segregating HTML & JavaScript. I.e. I was not for writing or generating HTML through JavaScript. Because, this will mean that the Front-end developers I have to hire will always be very strong on JavaScript alongside HTML5 & CSS. I was looking for a Framework that was on JavaScript but not HEAVY on JavaScript.

(3) The first iteration of the web app was to be done by myself. But I was clear that when someone takes up the mantle, they will be able to come up the curve fast.

In the end, Vue.js and Vuetify satisfied all the above criteria with aplomb! When I did our first POC on Vue.js I could not believe that front-end development could be this fast. The documentation was par excellence and all the required essentials that come along with the Framework (viz. Routing, Store, Validations) etc. were available from the same community! It was also a breeze to integrate with other JavaScript libraries (such as Amazon Cognito).

By picking Vuetify, we were able to provide a consistent UI experience between our Web App and Native App, besides making the UI development ultra blazing fast!

In the end, we were able to rollout our Web App in record 6 weeks (that included the end to end Loan Origination flow, Loans management system & Customer engagement module). www.jeyabalaji.com

Database is at the heart of any technology stack. It is no wonder we spend a lot of time choosing the right database before we dive deep into product building.

When we were faced with the question of what database to choose, we set the following criteria: The database must (1) Have a very high transaction throughput. We wanted to err on the side of "reads" but not on the "writes". (2) be flexible. I.e. be adaptive enough to take - in data variations. Since we are an early-stage start-up, not everything is set in stone. (3) Fast & easy to work with (4) Cloud Native. We did not want to spend our time in "ANY" infrastructure management.

Based on the above, we picked PostgreSQL and MongoDB for evaluation. We tried a few iterations on hardening the data model with PostgreSQL, but realised that we can move much faster by loosely defining the schema (with just a few fundamental principles intact).

Thus we switched to MongoDB. Before diving in, we validated a few core principles such as: (1) Transaction guarantee. Until 3.6, MongoDB supports Transaction guarantee at Document level. From 4.0 onwards, you can achieve transaction guarantee across multiple documents.

(2) Primary Keys & Indexing: Like any RDBMS, MongoDB supports unique keys & indexes to ensure data integrity & search ability

(3) Ability to join data across data sets: MongoDB offers a super-rich aggregate framework that enables one to filter and group data

(4) Concurrency handling: MongoDB offers specific operations (such as findOneAndUpdate), which when coupled with Optimistic Locking, can be used to achieve concurrency.

Above all, MongoDB offers a complete no-frills Cloud Database as a service - MongoDB Atlas. This kind of sealed the deal for us.

Looking back, choosing MongoDB with MongoDB Atlas was one of the best decisions we took and it is serving us well. My only gripe is that there must be a way to scale-up or scale-down the Atlas configuration at different parts of the day with minimal downtime.

Recently we were looking at a few robust and cost-effective ways of replicating the data that resides in our production MongoDB to a PostgreSQL database for data warehousing and business intelligence.

We set ourselves the following criteria for the optimal tool that would do this job:

• The data replication must be near real-time, yet it should NOT impact the production database
• The data replication must be horizontally scalable (based on the load), asynchronous & crash-resilient

Based on the above criteria, we selected the following tools to perform the end to end data replication:

We chose MongoDB Stitch for picking up the changes in the source database. It is the serverless platform from MongoDB. One of the services offered by MongoDB Stitch is Stitch Triggers. Using stitch triggers, you can execute a serverless function (in Node.js) in real time in response to changes in the database. When there are a lot of database changes, Stitch automatically "feeds forward" these changes through an asynchronous queue.

We chose Amazon SQS as the pipe / message backbone for communicating the changes from MongoDB to our own replication service. Interestingly enough, MongoDB stitch offers integration with AWS services.

In the Node.js function, we wrote minimal functionality to communicate the database changes (insert / update / delete / replace) to Amazon SQS.

Next we wrote a minimal micro-service in Python to listen to the message events on SQS, pickup the data payload & mirror the DB changes on to the target Data warehouse. We implemented source data to target data translation by modelling target table structures through SQLAlchemy . We deployed this micro-service as AWS Lambda with Zappa. With Zappa, deploying your services as event-driven & horizontally scalable Lambda service is dumb-easy.

In the end, we got to implement a highly scalable near realtime Change Data Replication service that "works" and deployed to production in a matter of few days!