Python Unconference

It's the first time I attended an unconference. In contrast to traditional conferences there are no talks announced upfront - talks are proposed at the beginning of the unconference, anybody can propose a talk, and the attendees vote for the talks they want to listen to.
This weekend the Python Unconference took place at the University of Hamburg - for three days, but I just attended Saturday.
26 talks were proposed for 12 planned slots (four sessions with three parallel tracks each). The proposed talks covered a broad range in terms of content and quality. After voting I attended the following talks:

• "Why Twitterbot? Using Python to Twitterbot" by Esther Seyffarth. The slides are available here. She introduced some of the Twitterbots she already has implemented mainly using Tweepy. Esther went through the code of OMG Wikipedia! in some more detail and showed various other Twitterbots - EmojiHaskell is my favorite one. Her motivation is that Twitterbots are a nice exercise for text generation using Python and can deliver some funny and entertaining results. In the discussion afterwards NLTK as an interesting toolkit for processing natural language in Python was mentioned.
• "TDD for APIs" by Michael Kuehne. To be honest I was distracted and didn't follow that much, but the discussion afterwards was focused on how to test the full stack of an API.
• The Lightning talks in between covered a broad variety of topics, e.g. 3D rendering of OpenStreetMap data and coding katas.
• "Pandas intro (Apache log analysis)" by Nikolay Koldunov. It was a live session by going through his IPython notebooks including an introduction into pandas and showing a use case for exploring Apache logs.
• "Building data products with Flask and AngularJS" by Andy Goldschmidt. He demoed two web applications - one used machine learning for classification and provides a simple interface to play around with the features of a data set. The other one analyzed an image to deliver the dominant colors. For both, the frontend was implemented using AngularJS, the backend is driven by Flask using scikit-learn for the machine learning algorithms.