All financial transactions are based on trust. In 1st world countries, banks can easily check your credit history helping decide whether you are trustworthy. However, in many developing nations, individuals do not have bank accounts and prefer cash transactions. Within tight-knit local communities, individuals know each other and are able to decide whether a person is trustworthy.
However, organizations/merchants outside of the community have no way of determining the trustworthiness of these individuals. Thus, many times, they are denied access to loans and other funding opportunities. Our goal is to use physical technology along with a machine learing model to solve this problem.
What it does
Our hardware solution enables people to build up a financial reputation while using cash in developing communities and provide micro-loans based on machine learning intelligence.
Our project consists of a RFID hardware hub which communicates with a central system. Whenever a transaction is made, the hub updates the central system with a record of the transaction. If the user does not have enough funds to cover the transfer, we automatically suggest a smart contract loan which takes their past and predictive future transactions in mind.
How we built it
CashFi includes both a hardware and software stack. We used an Arduino Uno with an RFID reader and tags to authenticate users. Then, we used Python with PySerial to parse the requests and send them the web server. The web server was built with Microsoft Azure. We used a Microsoft Azure virtual machine running Python Flask. In the backend, we use Scikit-Learn to run machine learning models to predict which loans to give out. Finally, we created a simple GUI using HTML.
Challenges we ran into
One of main challenges was maintaining privacy in a system with financial transactions.
Accomplishments that we're proud of
We're proud of our hardware + software stack solving a real world problem in an area that is typically shrouded in regulation.
What we learned
We learned a lot about developing both hardware and software together. We used multiple SDKs, APIs, and hardware components and it was a great experience.
What's next for CashFi
We want to develop a standalone embedded solution, so this can called installed in POS system and integrated with existing communities.