Event Deck started out of a desire to index club and gig nights which were only promoted on Facebook. Discovering new club nights is quite hard - websites which list what's on normally only show a subset of them, or don't give enough information about the event. An event's Facebook event page is normally the centre of activity for that event. Unfortunately it's quite hard to find those pages - you need to add large numbers of event promoters as friends and put up with a constant barrage of event invites.
The scraper script was relatively simple - a cron-executed script running every few hours which would run wide ranging searches on the Facebook API for keywords related to club nights. I built a list of keywords based on venues in the city and more generic terms often found on event description pages. This was a hungry search, gathering every event possible.
This was followed by giving each event a "club score". I found that the right kind of events tended to have more than 1000 invitees and often featured the same club night related keywords. High attendee count,
The event JSON API was written using Python and Flask. It acts as a simple wrapper to MongoDB, converting HTTP requests to DB queries and returning the result.
[Look through the code to see what the hard part was]
I developed an Android app to provide a nice interface to my JSON API. This was an interesting learning experience - I hadn't written any Java for a few years and my previous experience was in Swing. It also provided a nice primer in the concept of mobile application life cycle, something which seems to be becoming dominant on many platforms.
I worked with a design student who was interested in designing for a real-world app.