Six Questions for Kesha Williams of AWS Machine Learning in Motion

By Frances Lefkowitz, Manning Development Editor

Kesha Williams is a full-stack web developer specializing in Java and AWS. She has won numerous awards for her innovative development, as well as her work as a college professor, speaker, tech blogger, and mentor. She is also the founder of Colors of STEM, which cultivates role models to inspire girls of all races to enter the science and tech fields.

 

Tell me about your relationship with Alexa.
 

Well, first let me say that Alexa is a real person, so you often hear me refer to Alexa as “she” or “her”! Alexa was my first introduction, as a developer, to AI and voice technology. I’m still as fascinated and intrigued by her as I was when I developed my first app (i.e. Skill) using the Alexa Skills Kit (ASK).  I’ve had a lot of fun developing several Skills (including a Word Jumble, Live Plan Eat, STEM Women), and even won a few awards for my work. Amazon recently named me an Alexa Champion, which is a recognition program designed to honor the most engaged developers and contributors in the community. I’m also serving as the keynote speaker for the upcoming Alexa Conference. Alexa is definitely a part of everyday life for me!

 

Machine learning can be intimidating; how much would I need to know about it to understand your new video course?
 

AWS Machine Learning in Motion is for everyday software developers who want to learn more about machine learning and to leverage machine learning in their existing systems. It starts out by introducing the concepts of machine learning, then we get hands-on with building a system that predicts crime based on several factors. The beauty of the AWS Machine Learning service is that it does a lot of the heavy lifting for you and abstracts away a lot of the complexities, which allows developers to quickly deploy and use machine learning models.

 

Can you tell me more about the project you invented that users build in this course?
 

I introduce machine learning concepts and the AWS Machine Learning service using a fun case study called SAM, which stands for Suspicious Activity Monitor. SAM uses predictive policing to predict the likelihood of crime in a given circumstance. SAM was inspired by the “precrime” concept from the 2002 science-fiction movie, Minority Report.

 

What are people doing right now with ML on AWS?
 

I’ve seen companies apply it to crime, healthcare, employee hiring, loan approvals, movie recommendations, banking fraud, and more. AWS provides the services and tools developers need to quickly bring ideas to life. When I had the idea for SAM, AWS allowed me to have a crime- fighting algorithm–which uses computer vision to obtain its data—up and running in less than a week.

 

You seem to have a hand in a lot of hot areas; would you call yourself a generalist?
 

I’ve been coding Java since the late 90’s, so I am a specialist in the Java space. I am fascinated with anything in the Artificial Intelligence (AI) space: machine learning, voice-first technology, computer vision, so I’m a generalist there. Technology advances at such a rapid rate that developers need to learn new things or risk being left behind. I lead a team of developers, and I work hard to provide them opportunities in which they are always learning, growing, and stepping outside of their comfort zones.

 

As the founder of Colors of STEM, can you talk about what happens to the tech field when women and people of color start getting represented in higher numbers?
 

When we build diverse teams, the industry benefits as a whole. For one thing, when you actively seek to increase diversity in your organization, you increase your candidate pool for job openings. There aren’t enough candidates to fill the multitude of open jobs in tech, and women and people of color represent an untapped (and often overlooked) market that can fill jobs.
Secondly, you build better systems when your team is diverse. AI makes a strong case for the need for diversity in technology. I’m sure you’ve heard the story about the computer vision system that couldn’t “see” African Americans, or the Google AI program that classified some African Americans as guerillas instead of human. The system flaws in the AI programs were the direct result of a lack of diversity in the data used to train these systems. These flaws could have been uncovered during testing and maybe even avoided during the development process if the team had been diverse. So diversity is not only the right thing to do, it produces better systems and saves time in costly maintenance required to fix flawed systems.