Don Kent

Don Kent

Data Science Manager, Intel

Meg Hermes

Meg Hermes

User Reference Manager, JMP

Don Kent works for Intel Corporation’s Technology Development Group as manager of a full-stack data science team. Prior to joining Intel, he served as Principal Data Scientist at IMFlash Technologies, a joint venture between Intel and Micron. 

With his background in engineering and solid-state physics, Don has been a part of the semiconductor industry for two decades and a JMP user for nearly as long. An outspoken advocate for analytics for many years, he helped found the Wasatch Front JMP Users’ Group and co-hosted countless JMP meetups at IMFlash.

Meg: You’ve been a JMP user for almost 20 years! In that time, you must have seen quite a few changes not only in the software but also in the field of industrial statistics more broadly. How do you think your personal analytics journey ultimately shaped your career path?

Don: My advocacy for analytics has largely paralleled my journey of career growth. Looking back, it started with a keen interest in learning more about statistics. As a novice, the challenges were at the personal level, focused around finding mentors and making connections with like-minded folks. Much of the time in my earlier years was spent with user groups or other communities, both in contributing and in learning as much as possible.

It’s interesting how, as you look back, you can see a few key milestones that stand out. For me, those milestones are related to someone encouraging me to learn and to grow my skills – either from one of the amazing mentors I’ve known or, for example, from a manager who encouraged my journey with a complete boxed-set of JMP 4 books (which are now PDFs!).

As my knowledge, responsibilities and confidence in statistical methods have grown, so too have the opportunities I’ve had to impact the organizations I worked in. I see a clear correlation between my analytic growth and the career opportunities I’ve been afforded.

I started my career with Intel as a reliability engineer with a background in semiconductor device physics. However, through my interest and study of analytics over the last 20 years, I've been able to transition my career to managing a data science team within one of Intel's Technology Development groups. That opportunity to both create and lead an analytics team would not have been possible without the mentorship of others and the confidence of my current manager.

As I've grown in skills and knowledge, the dynamics have shifted from personal advancement to more interpersonal advancement. Now I find more rewards in mentoring and encouraging others in their journeys.

Meg: Yes, mentorship is absolutely key – no matter what role or industry we’re in, we can always learn from those who have walked the path before us and encountered similar obstacles along the way. Now that you’re in the position of mentoring others, what would you say are some of the primary hurdles you’ve learned from?

Don: One that comes to mind is related to spatial pattern analysis. At the time, I was working on a project at a former employer, and I remember going to my manager and showing him a JMP training course called "High Dimension Data Analysis." I explained how I thought that course would significantly increase my knowledge in this area and help with the project. Unfortunately however, my manager highlighted that at the time the company did not have the budget for the training.  

I was of course dismayed, but after some consideration I realized that my education was in my own hands. So I paid for the course myself and took the class. It was one of the best courses I've ever taken, and it significantly helped that project. Even more importantly, coming to that decision was a real turning point in my career. It helped me focus in on what was important and helped bring clarity to the direction I ultimately wanted my career to go.

Meg: That provides me with a nice segue to ask you about something we often don’t talk about much, which is the wealth of resources JMP offers and the hidden value it brings to our customers. Can you speak a bit about how you’ve leveraged those learning resources and opportunities to interface with the JMP organization over the years?

Don: JMP really has the entire ecosystem to enable an individual to get involved at many different levels. I personally have utilized so many of the features of the JMP ecosystem that it’s hard to really overestimate the value, so to be brief I will mention one.

For those who have never attended a Discovery Summit – I just can't recommend it enough! The JMP team actually feels like a family, and they bring you into that family, which is a pretty amazing experience. This is all on full display at Discovery. Every time I’ve attended the Summit, I’ve taken home not one thing but many things, and I consider it a key conference. The access to tutorials, trainings and presentations is amazing.

The JMP team also offers opportunities to visit with the developers directly. They love to hear from customers and understand how they can make the product just a bit better. Of course, the keynotes are always by current thought leaders, and I personally treasure that one dinner sitting and visiting with Stu Hunter as one of the highlights of my many wonderful Discovery experiences.

Meg: That is so wonderful to hear! Speaking from the perspective of someone inside of the JMP organization, I have to say that I too thoroughly enjoy Discovery Summits. They are a real highlight of my job.

One of the most common objections we hear from newcomers to JMP is that some fear statistical approaches will supplant domain expertise. How has having JMP changed the way you and your colleagues apply domain knowledge and skills?

Don: Within our organization, the Technology Development team is eager to utilize JMP for many reasons – not least of which is that it significantly helps them complete their tasks. In fact, I would venture to say that the concern from our side is that the team needs to combine their expert knowledge of the semiconductor process with the correct JMP platform (and statistics) to answer a desired question.

Due to certain semiconductor hierarchies – for example, die within wafer within lot – certain data analysis can go a bit sideways if the proper hierarchy (or expert knowledge) is not correctly utilized. For example, using the proper lot-wafer-die nesting structure in a variance components analysis can make a large difference in the result.

Also in data mining. Choosing full lots (all wafers) for training and/or validation data sets will reduce the likelihood of information leakage when data mining at wafer level. We have to understand these complexities and combine expert knowledge with the correct JMP platform and statistical approach. As our statisticians like to say, "When in doubt, consult your local statistician."

Meg: Speaking as that statistician or go-to person for analytics support, what advice would you give to someone looking to cultivate a more mature analytics culture in their organization?

Don: What does a mature analytics culture look like? For us, we look at the culture across a few vectors – and the key is the team – the individuals in the organization must have the analytics knowledge or skills to enable them to perform their jobs with confidence and efficiency. This requires training and the knowledge to best use the tools they have, tools like JMP. They don't need to know everything. And that is where the analytics team can fill in the gaps.

Another highlight of a mature organization is the availability of KPIs and metrics for the organization. These should be standardized, available and visible across the enterprise to help direct the entire team. Finally, look for advanced applications that involve machine learning or utilize recommender engines – organizations that are utilizing these advanced tools have evolved data ecosystems, whereas organizations that are not quite there should work on improving the fundamentals like data integration, pattern/feature identification, automation and predictive analytic systems. Of course, having a dedicated team can significantly help drive these initiatives.

Meg: Speaking of new initiatives, introducing new analytics workflows can be disruptive at first. But, as we know, the benefits of an effective analytics strategy outweigh the initial costs. What kinds of ROI or improvements have helped to build and sustain momentum for data initiatives at Intel?

Don: In terms of adding value to the organization, the first step is generally a gap analysis – understanding where you, your team and the organization currently are compared to the direction they should travel. “It's a journey not a destination” can be a tremendous motivator and shine the light on where improvements and efforts will have the most value and help define the ROI.

Identifying those gaps can be time-consuming but, in my experience, involves meeting with and listening to customers and business partners. As an analytic organization, it is our job to translate the business need into a data strategy or data problem. By engaging and listening to our customers and partners, we are empowered to work with the organization to help synthesize the best analytic strategy. Once that strategy is determined, then executing on it becomes the focus. This is where engineers and statisticians excel and can shine.

One motivator that builds momentum is progress toward goals. Simple things like successfully deploying an add-in or adding a feature that was requested by a customer builds trust and progresses the organization’s goals. Often to successfully deploy these features requires partnering with other teams like the training group, who can help with professional documentation and communication. Metrics can also be powerful tools to gauge momentum and adoption and are used extensively to help understand and sometimes shape customer engagement.   

Meg: As we wrap up, what advice would you give to someone who is just starting out with JMP?

Don: Be curious, use the "Help" section of JMP – and do so frequently! The documentation within JMP itself is the best place to begin, and the sample data sets and books are a treasure. From there, check out the resources online at, check the training schedule and get engaged with a users’ group – try virtual if in-person is not available in your area.