I haven’t posted for a while because I was getting ready for my first tech conference talk, which I gave at PyCon 2017 in Portland last month.

My talk was about an angle of A/B tests that I hadn’t seen before: in most talks on A/B Tests, there’s this black box that “randomly assigns visitors to see either A or B”. At Yelp, I led a small team that implemented the experiment assignment library. I thought I was in a good position to explain what could go on in that black box (using Python) and why it mattered to everyone else running experiments.

Watch it here:

And here are the Slides on Speaker Deck.

And a few Jupyter Notebooks with more info.

Other PyCon talks

It was my first talk at a tech conference, and the Green Room was great. I went there and nervous-chatted with a bunch of folks, including David Wolever, who organized the Green Room! (Awesome!) and then gave a cool talk on disassembling Python code, and Michelle Fullwood, who gave an informative talk on TensorFlow!

While I’m at it, I also learned a lot about things I’ve been meaning to look into, like stream processing by some fellow Yelpers. One Data Pipeline to Rule them All is a nice overview data pipelines, which could have been useful last year before I started contributing to Yelp’s data pipeline.