Getty Images - The Feedâ„¢ while at R/GA
We designed and wrote an application in python that listens to the twitter streaming API. It works out when it
needs to search for images and what to search for. It ran during the superbowl and handled around 1mil tweets.
It has been patented and it has won awards.
If I ever work on a better project, with better team-mates, for a better client, I've won the internet.
White-board Sessions!
Some of the most fun and innovative programming and process techniques went into this project. Here we are
discussing how the Social Listenerâ„¢ (patented) will post (and award winning) pictures in real-time during events
like the Superbowl.
Simulated Testing
This a super early version of the system listening to a simulated stream of tweets coming out of MongoDB.
The data is a Champions League game between Barcelona and Celtic.
Across the bottom the tweets are recorded by a Listener and their keyword scores are saved into Redis.
Then right to left across the top a Monitor is keeping track of active searches in the system and when it should
have a look at the data. It fires signals to an Analyzer that will create tables of keyword scoreboards and then
decide what to search for. Tasks are then sent to Gearman to search an image database and post the results to
various different social channels.