How Companies Do Machine Learning

I came across a post from Pinterest’s recent acquisition (hat tip OReilly’s Data Newsletter) and it made me wonder how some of the “cool kids” in e-commerce and other online properties use machine learning.  Surprisingly, it’s kind of scattered and hard to find.

Here are just a handful of blog posts and videos I was able to find on companies like Google, Netflix, Pinterest, OkCupid, and Etsy.


Google (per Edmond Lau on Quora) doesn’t use much machine learning for their organic search ranking.  Instead, they use hard rules to be able to control their search results better.  I always loved this Search Quality Meeting video (YouTube) that Google posted a couple years ago.  Definitely interesting to see how Google’s main product is maintained.

A little bit more on the technical side, Andrew Ng talks about his research on deep learning at Google ( for image recognition.


Netflix has their TechBlog but one of my favorite posts is their insight in to the Netflix Challenge results (part 1, part 2).

Secondarily, Xavier Amatriain, formerly the Research / Engineering director of Netflix, has great answers on Quora (where he is now VP of Engineering) and blogs occasionally (blogspot) as well.


Pinterest, with its recent acquisition of Kosei, listed some of the machine learning tasks (and their team names).


Etsy has a Code as Craft site which they post fairly long, technical posts.  They shared some insights on their recommendation system (CodeAsCraft blog) in late 2014.  Has some interesting insights in how they’re using MapReduce and Recommender Systems.


I’ve posted about AirBnB’s engineering blog under a post about making analyst’s more effective. AirBnB has an amazing system set up for sharing knowledge internally and I appreciate them sharing their successes and challenges with the rest of us.

The topics on their medium blog range from programming to machine learning to corporate culture. Highly worth subscribing to.


Lastly, OkCupid has mainly statistical analysis posts on their blog.  I love the insights with pretty simple statistics.  The author also wrote a book (Amazon) with some mixed reviews.

Everyone seems to write about business successes (Amazon, Zappos, Google, Good to Great, 3G Capital, etc.) but there aren’t that many books extolling the success of data mining / machine learning written by practitioners.  It’s too bad really.  There’s obviously a great deal of knowledgeable people out there.

Maybe just a few more years and we’ll have an anthology of advice articles on building real-world machine learning systems.