Hackathon at Oxyma

As an omni-channel agency we are not new to hackathons. Last year our aFrogleap team spent a week at SXSW in Austin for winning the Liberty Global appathon. But we were new to organizing Hacktathons ourselves. Up until last friday that is… Our Rotterdam office hosted the first Oxyma Exploration Days Hackathon. Attending teams showed off their skills and hacked their way into bots and AI’s using a combination of coding-, data-, & creative skills.

We started off at 8 AM with a good healthy breakfast from our downstairs neighbours New Fork (they make mean yoghurts and incredible croissants). Our group CEO Jos van Loo welcomed attendees with an over the mountaintop welcome message and attendees shared their guilty pleasures before booting up their Raspberry pi’s and programming gear. In our setup we combined programmers, data scientists, creative and concepting skills in each group in order to get the best results.

During lunch every team presented their concepts for bots and AI. All afternoon and evening the teams worked on their ideas. Idle time for creatives and concepters was spent on a parallel track creating new solutions for bra-fit challenges for online buyers of Marlies Dekkers lingerie. After loads of Red Bull, Chupa Chups, Brabantse worstenbroodjes, healthy food like pizza and an occasional beer, attendees presented their concepts in a late night science fair. To give you an impression we’ve listed the ideas below:

What’s for dinner

This voice controlled Amazon Alexa skill takes the contents of your fridge and suggests a dinner recipe based on ingredients that have the shortest time to expire. It does not just tells you what to cook but also sends you on errands for things missing in the fridge.

Code on github

Please (don’t) go

Ever been to a party only to find out the crowd isn’t your thing? Well, no more! This app lets you scan a parties attendees beforehand. Based on your preferences and attendee Facebook profiles this app mixes and matches parties and venues you should attend.

Code on github


This is truly an AI piece of art. The Jukebot team hooked up a raspberry pi to a camera and created an algorithm that scanned the room and counted the number of people and their activity level. Useless you say? Ha! They actually used crowd activity to modify the music in the room. How’s that for crowd control!

Code on github

Want to attend upcoming exploration days? Stay tuned!

No comments.