Using Experiment Trackers for Machine Learning Models

Hugo speaks with Hamel Husain, Head of Data Science at Outerbounds, with extensive experience in data science consulting, at DataRobot, Airbnb, and Github.

In this conversation, they talk about Hamel's early days in data science, consulting for a wide array of companies, such as Crocs, restaurants, and casinos in Las Vegas, diving into what data science even looked like in 2005 and how you could think about delivering business value using data and analytics back then.

They talk about his trajectory in moving to data science and machine learning in Silicon Valley, what his expectations were, and what he actually found there.

They then take a dive into AutoML, discussing what should be automated in Machine learning and what shouldn’t. They talk about software engineering best practices and what aspects it would be useful for data scientists to know about.

They also got to talk about the importance of literate programming, notebooks, and documentation in data science and ML. All this and more!

Links

• Hamel on twitter
• The Outerbounds documentation project repo
• Practical Advice for R in Production
• nbdev: Create delightful python projects using Jupyter Notebooks