CredLo was inspired by a personal story of a teammate that talks about the various challenges that immigrants face when moving to a new country. The primary challenge among immigrants is restarting their life in a new place, which begins with their inability to obtain credit as credit scores are non transferable across countries. On top of this, we saw many individuals lack access to a lump sum of money quickly at low interest rates, sparking the need for an automated micro loan system that is diversified, low-risk and easy to use.
What it does
CredLo uses user-inputted data and personal submissions to generate a credit score for the country that the individual is moving to. Additionally, borrowers are able to attain loans quickly at low-interest rates and lenders are able to lend small amounts of money to a large number of people people at a level of risk that they choose.
How we built it
We built the backend using Flask/Python to process requests from the lender/borrower as well as for the borrower's onboarding process. We used Capital One's API to make actual transactions between the lender and the lendee. We trained our ML model using sklearn on a dataset we found online. Most of the frontend was built using vanilla HTML/CSS/JS (no wonder it took us ages to build the UI), with a little bit of Vue sprinkled in. The data was stored as a JSON object (with periodic serialization, which to answer your unasked question, yes we eventually intend to use Cloud Firestore for this instead :)
Challenges we ran into
- Naming the product and coming up with the tagline was difficult
- Since none of our teammates are front-end developers, a large chunk of time was spent by trying to make our UI look somewhat bearable to vue (expect a few more puns as you read along). Time spent working on the UI could have been spent working on additional features instead.
Accomplishments that we're proud of
- As a team with zero front-end developers, we have a passably pretty UI.
- We are proud that our product attempts to solve a real need posed by many individuals around the globe. We had other ideas that were more technically sophisticated, but we instead decided to work on a product that had a real-world impact and could positively impact lives in society. After speaking to various individuals in our target market who said that they would have greatly appreciated assistance from the CredLo platform when moving countries, we are proud that we developed a product that can be incorporated into society.
What we learned
We learned about the various challenges that different groups in society face and the ways in which we can alleviate their stress and headache. We also learned to collaborate and work together as we are a group of students with different backgrounds and skills.
What's next for CredLo
- We were restricted by the kind of datasets we had available to use to generate the credit scores. With more time and research, we can improve on the metrics used to come up with an accurate credit score. The eventual goal is to work with banks and other institutions to become a reliable source of information that individuals and institutions can trust.
- Instead of using user input (which can be faked), we would include verifiable sources of claims such as bank statements, utility bills etc. and extract the necessary data out of them using computer vision.
- There is currently only an auto-investment mode for lenders. That is, they do not choose who they can lend their money to. We would like to expand the project to allow investors to choose people they think have a sincere need, adjust their rate of interest down if they so wish to, along with the amount of investment (up or down). Eventually, CredLo would provide lenders the possibility to manually invest their money instead of having it automated.
- Complete integration with Capital One's APIs to facilitate actual bank transfers. We started working on this but left it unfinished due to technical issues.