Beyond Hackathon 2017

This weekend I participated on Beyond Hackathon, the 8th (and biggest) Hackathon I’ve joined with friends from DSG.

About the event

Beyond Hackathon was a FinTech Hackathon, organized by Eurobank between 17/03 - 19/03 at one of their buildings. The event attracted 38 teams from more than 5 countries. The first day was reserved for introductions/pitching so we only had the weekend to work on our ideas.

Pros

• Nice prizes for the first 3 teams, 5.000€, 3.000€ and 2.000€ accordingly
• The place was great
• The food was acceptable(considering they had to feed more than 100 people)
• Almost uninterrupted high speed internet thanks to the technical team
• Helpful and available mentors (we only talked to ~4)
• Godlike host (Stavros Messinis)
• A lot of participants, many of them from other countries

Cons

• We couldn’t spent the night there in order to keep working
• Teams had too much time(5mins) for the final presentation, thus more than 3 hours for everyone to present

Our team

Our team consisted of 3 people, Dimitris, John and me. Due to difficulties Dimitris managed to join us only for the last day of the hackathon but his help and ideas were very valuable.

The idea

We had almost no experience around FinTech so we went without any idea on what we are going to build(like most hackathons in the past). After hearing the first mentor talking about customer experience we decided to build a system that will use facial recognition in order to inform the bank’s staff when a high value customer enters the branch so they can help him faster.

Our implementation

We had to code the following:

• A simple mobile app that would simulate the camera at the branch’s door so we could take the customer’s photo
• Some sort of backend that would use the Microsoft Cognitive Services - Face API and connect all the parts of the system.
• A webapp for the bank’s clerk that would inform him when a high value customer entered the branch and show him some information about him(name, last transactions, account balance etc)
• A way to greet the identified customer via his mobile phone(sms, app, chatbot etc)

Day one

I spent most of the morning trying to generate azure keys and figuring out how Microsoft’s Face API works. While I was battling Azure, John was writing the mobile app that would simulate the bank’s camera. After the first lunch break we had a trained model with 3 faces and the camera app ready.
Sending the photograph to the backend and then checking it for a match seemed like unnecessary complexity so we decided we would use the Face API from the mobile app and just inform the backend on a successful match.
John started coding the API calls for the mobile app and I started working on the backend/frontend for our web application that the clerk would see.
By the end of the day we had a prototype! We could take and analyze photographs from the mobile app, forward successful matches on the backend that would then update our 20 line webapp via sockets in order to see who “entered the bank”.

Day two

The next morning we started thinking about how we would greet the customer that just got identified. After talking with the mentors we decided that on a real scenario the user could be informed by eurobank’s mobile app, since we didn’t have the options for that during the hackathon we decided to create a dummy application that would demonstrate the concept.
For the rest of the day I wrote code for the dummy customer app and got some data from Eurobank’s API while John was finishing the web application. Around noon Dimitris arrived and started working on our presentation.

All of the source code can be found on our GitHub repo

Presentations and winners announcement

The big time came, we didn’t have much time to prepare for our presentation but thanks to previous experience, especially on Dimitris’ behalf we didn’t do bad. We demonstrated our system by taking Stavros Messinis’ photograph on stage and identifying him as a high value customer(with a delay of ~2sec that seemed like an hour).

The results

A few hours later the judges had decided. We didn’t get any of the first places but we were pretty happy because once again we succeeded on creating a working MVP in very limited time and we experimented with a pretty nice technology for the first time.

1st place - Block17

The first prize went to Block17 for their Blockchain-based bank-to-bank settlements system.

2nd place - Hacking squad

Hacking squad, a team from Serbia got the 2nd place for implementing banking functionalities with Amazon’s Alexa.

3rd place - Wire Buzz

Wire Buzz got the 3rd place for an automated wire transfer notification system.

Closing

Overall the event was great and very well organized. I met a lot of great people and saw familiar faces from other hackathons. Hopefully I will be able to participate again next year!