Driven by his passion for transparent Al, data science consulting, and ML education, Keith McCormick, joins the Pandata team as a data science executive in residence.
His 30-year background in statistics and ML modeling will not only benefit Pandata’s group of data scientists, but also add tremendous value to existing and future client partnerships. Read on to learn more about Keith’s journey and his plans to impact the future of Pandata.
Looking back, I realize that I started coding at a young age. The scale was nothing compared to what I did in college, or to what I do now, but I did show an early interest in the field. I went on to attend Worcester Polytechnic Institute (WPI) in Massachusetts on an Army scholarship. I studied computer science and became increasingly interested in psychology—which comes in handy when you’re solving problems for people.
After receiving my bachelor’s degree in computer science and psychology, I wanted to continue my education with a Ph.D. and moved to North Carolina to do so. However, while establishing my state residency, I had the opportunity to become an external software trainer for SPSS Inc., the company that made the well known statistical software. . I thought SPSS would only be a side hustle while pursuing my Ph.D., but it turned into a career, eventually teaching hundreds of SPSS classes to thousands of students, many of whom were working PhDs in academic, government, and industry.
I was very active in this space from the late 1990s until IBM bought SPSS in 2009. Within IBM consulting was valued more than training, and SPSS Modeler (the machine learning software) was valued more than SPSS Statistics. I believe this was a career milestone that shifted me more towards consulting and machine learning (ML).
I originally connected with Pandata’s CEO, Cal, at ODSC in Boston in early 2022 and quickly realized a mutual passion for transparency in AI. Shortly after the conference, I was approached by a client to do an interesting project but didn’t have the bandwidth to take on a new analytics project. I reached out to Pandata and referred the work to their team. As the work went on, Cal encouraged me to sit in on update calls to closely follow the progress, mentor the team, and stay connected with the client. Overall, it was an amazing experience.
With the success of this project, Cal and I considered implementing the same process on an ongoing basis and arrived at the partnership we have today. In this new role, my core responsibilities will be mentoring Pandata’s data science consultants, supporting business development efforts, being a trusted advisor to client executives, and evangelizing responsible AI.
There is so much skill already on the Pandata team. I believe that my experience in consulting is going to be a great addition to the team. A former boss once said, “Data scientists are like 777 pilots; by the time you’re any good, you’re close to retirement.” In most cases, data scientists require three to five years of experience before they’re a project lead. Even after 10 years of experience, a data scientist may have only led two or three major projects.
As a mentor, you can find yourself mentoring up to 12 projects at a time. For a project lead, this volume would be inconceivable. After 10-15 years in the data science and AI industry, you’ve had enough experience to transition into mentoring roles. This is where I’ve been in the last 10 years of my career and with Pandata, it’s allowed me to participate in more projects than if I was still focused on just a couple of projects at a time.
I noticed that the topics Pandata emphasizes—data privacy, trustworthy AI, risk mitigation, bias prevention—seem to be trending and resonating with a lot of ODSC attendees. A number of people who have approached Pandata agreed that more practitioners and companies alike need to talk about AI ethics.
ML operations is also a hot topic, and although it may not be an active space for Pandata, it’s important to know. Machine learning lands between data engineering on the front end and ML ops on the back end. As data scientists and consultants, we need to understand what the vendors are doing on both sides: The models we build will be using data provided by data engineers and may eventually find their way into systems built by ML ops professionals.
Right now, I’m reading “Competing on Thought Leadership” by Robert Buday. The book lays a strong foundation for the reasons why companies need to be disciplined about thought leadership strategy. I also love that it helps differentiate between product marketing and thought leadership marketing. Since I’m a SME, and don’t have a marketing background it’s cleared up a lot of things for me. It’s a good read.
A great piece of ML advice that has stuck with me is that you need to think about who the end users will be and how they use the model before it’s built.
A great metaphor for this is a blinking light on a hospital floor and being the nurse that has to run down the hall to check why the blinking light came on. That’s what most ML models are—determining risk, and when there is elevated risk, someone has to act and execute an intervention strategy. When the model is deployed it might actually put new demands on these end users, and we have to respect their role in the process.
We tend to focus on the the internal client, but the end users are the real consumers of AI. If they don’t trust it, they will derail the project—unless you earn their trust, they won’t use the model. This mindset is at the forefront of every ML project I consult on.
Something I do now, but didn’t do when I first started consulting, is treat the initial client call almost as if it were a job interview. In other words, do diligent research. Start by looking up the organizational chart, team backgrounds, company history, historical revenue, and so on.
The first time I handled a consulting project on my own, I wasn’t sure if I added the total project hours correctly. I then started to worry about the cost of the project. The issue? I was looking at the numbers through my lens, not the client’s lens. A mentor of mine suggested looking up the company’s financials. They were a multi-billion dollar company. Upon doing so, I realized the cost wasn’t a problem if I could demonstrate the potential value of the project. It never occurred to me to look up the balance sheet or annual report of the company that I was working with.
Our team of data scientists, including Keith, regularly contribute their insight to Pandata’s Voices of Trusted AI email digest. It’s a once-per-month email that contains helpful trusted AI resources, reputable information, and actionable tips.
Contributor: Nicole McCaffrey is the Chief People & Marketing Officer at Pandata.