Chelsea Troy is a self-taught and informally educated software engineer and data scientist who also specializes in machine learning. Chelsea also blogs regularly at www.chelseatroy.com. In this conversation, we discuss the process of self-educating in a variety of software-related disciplines, the state of machine learning and whether or not its going to swallow our society, and how technology companies can improve diversity in their workforces - both in terms of tangible actions for employees and managers as well as higher-level organizational changes.

Check out more from Chelsea here:

• Instagram: @misschelseatroy
• Website: www.chelseatroy.com

If you're enjoying the show, the best way to support it is by sharing with your friends. If you don't have any friends, why not a leave a review? It makes a difference in terms of other people finding the show.

You can also subscribe to receive my e-mail newsletter at www.toddnief.com. Most of my writing never makes it to the blog, so get on that list.

Show Notes:

• [0:07] Self-educating in software development and data science through a project-based approach – and the strengths and weaknesses of project-based learning vs a formal academic model
• [08:48] Almost all of your time in software development is spent at the margin of what you know how to do, so you have to be comfortable with being uncomfortable. Improvement often comes through bettering your ability to solve the inevitable problems that you will run into.
• [19:12] Reduce the feedback loop as much as possible and create testing scenarios in order to rapidly iterate on software. One weird trick to learning software development: copy the changes that more experienced developers make to their code by hand
• [30:30] The best learning comes from realizing that you’ve made a mistake. Having a generalist approach and understanding multiple programming languages enables solving problems in non-traditional ways.
• [37:42] Should we believe the hype on machine learning? What will be the future of machine learning and how will humans work with this technology as we are able to automate more and more tasks and better recognize patterns in data?
• [48:02] The dangers of algorithmic recommendations and the amount of resources going into increasing advertisement clicks through machine learning. Can we have machine learning algorithms make their decisions and categorizations “human legible”?
• [1:03:07] How can tech companies move the needle on diversity in hiring? What actionable communication and management behaviors can individuals employ in terms of making technical companies more welcoming to underrepresented folks?
• [1:14:07] How do we get more viewpoint diversity in the upper echelons of technology companies? Viewpoint diversity seems to clearly help companies improve performance, but can be painful and create more conflict within the organization.

Links and Resources Mentioned

• John Conway's Game of Life
• Deliberate practice
• What is the difference between FragmentPagerAdapter and FragmentStatePagerAdapter?
• GitHub
• Pivotal Labs
• “Leveling Up Skill #6: Commit Tracing” from Chelsea Troy
• Zooniverse
• Hubble Telescope
• Hanny's Voorwerp
• Janelle Shane
• “Try these neural network-generated recipes” from Janelle Shane
• “Do neural nets dream of electric sheep?” from Janelle Shane
• “Metal band names invented by neural network” from Janelle Shane
• “The neural network has weird ideas about what humans like to eat” from Janelle Shane (this one kills me)
• Decision tree learning
• Game of Thrones