Thank you so much for joining us in this interview series. Before we dive into our discussion our readers would love to “get to know you” a bit better. Can you share with us the backstory about what brought you to your specific career path?
I found my passion for digital enablement on a pretty unconventional path. I didn’t begin my career as a data scientist, or in advanced technology — I began with the US Marine Corps.
Through my deployments around the world, specifically in sub-Saharan Africa, I witnessed what happens to communities when they lack access to health care and have unmet medical needs. What I witnessed stuck with me until after I left the service, so when my wife began her Ph.D. program in Biochemistry at Duke, I decided to continue my education in order to support the larger medical advancements and help underserved communities across the globe. I spent those years learning all about biotechnology, specifically biopharma, which is industrial fermentation, where we use cell machinery to make a product.
After receiving my degree in bioprocessing science, I began working at Genentech as a process development engineer — focusing on scaling up pharmaceutical production to commercial level, then transitioned into manufacturing sciences as a process engineer in Good Manufacturing Practices (GMP) production and later as a reliability engineer. In both positions, I really nerded out on the impact of people and equipment uptime, how they work, how it breaks down, and how those people and pieces work together. That whole part of my career was focused on leveraging data to get answers. After this experience and a short stint on my own as a business owner, I landed my current position at CAI, where we work to ensure a company’s pharmaceutical facility is up and running on time, on budget, and performing as expected and designed.
At CAI, I work with many different companies, seeing how a lot of the industry manages assets, reliability, and most importantly, what good and bad data looks like. I learned the pain points of those involved in drug development and manufacturing and slowly began to think of how to improve predictive maintenance.
Predictive maintenance tools in pharma manufacturing operations that I began using ten years ago have now incorporated industrial internet of things (IIoT) to enrich data and artificial intelligence (AI) and today’s computational power to translate this enriched data using complex modeling evolving how we predict performance and more importantly, predict failure.
What do you think makes your company stand out? Can you share a story?
When the construction of a drug manufacturing plant is complete and all the needed equipment is inside, the facility is still not ready to produce and deliver safe, effective treatments to patients. There are many remaining tasks to ensure operational readiness — transferring the technology and product into the plant, testing the plant and processes, standing up programs in asset and quality management, and ensuring the workforce is set up for success — the list goes on! Guiding companies through these processes and the biggest struggles of the pharmaceutical industry is the crux of what we do at CAI.
Number one is drug shortages. If companies are not managing assets, both tangible and intangible, and programs in a safe, reliable manner, drug production, and delivery are delayed, leading to a lack of steady supply for patients whose lives depend on them.
An additional industry issue is bringing the latest developments in personalized medicine to market in a cost-effective manner. When drugs are personalized, they can’t be mass-produced and therefore are expensive to manufacture. Bringing down the overall manufacturing costs is important, and our team provides individual services to support that goal.
You are a successful business leader. Which three character traits do you think were most instrumental to your success? Can you please share a story or example for each?
Number one: An entrepreneurial mindset.
A person with a good idea doesn’t always create a good solution. The person will have to translate that idea into a solution and gather a team to bring it to fruition.
For example, my role in digital enablement is very new for the company, but it’s also fairly new for the industry. While some guidelines have been given, the “How to induct digital enablement in drug development and manufacturing” hasn’t been established. Action is better than waiting for the right solution to present itself, and we’re driving innovation to implement the latest technological advancements into processes. An example of this is our collaboration with a large pharma client last year. The team was shifting their facilities process from one system to another and wanted to ensure their underlying facility data was correct. Therefore, they asked our team to completely rebuild the facility data structure from scratch within 75 days — despite the fact our team was located in a different state.
With a local team using reality capture in person at the original facility, we were then able to collaborate globally with a much larger remote team to build the digital twin. Using this digital twin technology, any remote team member could walk that facility as if they were there. The introduction of advanced technologies allows us to act globally but affect locally.
Number two. Be a lifelong learner.
Every professional role I’ve had is connected to data and digital enablement. One of the reasons I made this transition is my fascination with the latest industry advancements. AI, augmented reality, and digital twinning are happening now, and they will play a crucial part in the industry’s growth over the coming decades.
Lifelong learning does not always mean going back to school to earn another degree. I want to make sure I remain knowledgeable and competitive — prioritizing discussions with industry thought leaders and always reading up on the latest technologies. Some of my greatest mentors have tackled new and complex technology and concepts on the verge of retirement, demonstrating their commitment to learning, and I’ve always respected that approach.
Number three: Support the education of those around you.
When I spoke with pharmaceutical engineering thought leaders about the pain points of AI, many were concerned about very technical aspects of AI integration — how we should deal with regulatory agencies and how we should monitor, track, and validate models. That was what they believed were the most important areas of education for AI. These are important topics that need consideration and thought leadership, but I have a different priority in education.
While those are the concerns of ten AI experts, there are tens of thousands of people within the wider industry who don’t know the basics of how AI will be involved in their work. To ensure the full adoption of breakthrough technologies, we need to demystify AI and make it less intimidating for beginners. A rising tide lifts all boats and when all levels of our industry improve their data acuity and familiarity with AI, we all benefit.
For example, I am working to coordinate a special hackathon to feature companies that have developed democratized AI solutions that don’t require advanced coding skills. I want to emphasize AI solutions that a layman can use, putting a problem in front of a group and saying, “Here are your tools, go find the solution.” Our scientists and engineers in the industry are some of the brightest in the world and with the right tools, they can solve our most complex problems.
Every person within the industry should have a basic knowledge of data acumen and AI. Otherwise, the widespread fear will continue.
Let’s now move to the main point of our discussion about AI. Can you explain how AI is disrupting your industry? Is this disruption hurting or helping your bottom line?
AI and machine learning models have actually been involved in the research and development side of the pharmaceutical industry for quite some time — a good friend of mine began working as a computational biologist all the way back in 2010. Developing an effective COVID-19 vaccine as quickly as Moderna and Pfizer and BioNTech did would not have been possible without the use of machine learning models.
The presence of AI supported great advancements within the clinical trial stage. The patient subpopulation chosen for a clinical trial can make or break a therapy. If the selected patients within the clinical trials improve from the treatment, it can then progress to future approvals, and then to market.
Alternatively, if a drug fails in clinical trials, there is a high chance the drug is declared fully ineffective. In reality, they may just be ineffective in that subpopulation and could potentially improve other subpopulations’ quality of life.
The cost to bring a biologic to market is almost $2 billion, and half of that money is spent before Phase Three clinical trials even begin. Now, diagnostics companies are utilizing digital twin technology to help companies better select patients for clinical trials, reducing the chance of wasted costs.
In addition, we are in our infancy of utilizing AI to aid in GMP pharmaceutical production and supply chain management. Given it is a highly regulated environment, the integration of AI in the production phase is further behind the research and development phase.
While there is work going into the validation aspect and the use of AI in the GMP space, I believe AI will first be incorporated into augmentation tools in order to automate tasks that are not seen as valuable to the employee but where humans are significantly involved. If we have a similar rigor to data integrity and governance, we’ll build a solid foundation for use in the validation space. At CAI, we’re breaking down workflows in how we deliver work and automate many of these tasks using large language models and computer vision — all while establishing the fundamentals of a modern data architecture to enable our teams with better tools. In parallel, we are developing a framework for model development and deployment for testing and validation of machine learning models for future use. This hyperfocus on work optimization helps the organization defend against potential disruption from outside players while future development allows us to play offense in disrupting the operational model for operational readiness and speed to patient.
Which specific AI technology has had the most significant impact on your industry?
It is the technology that can be widely understood and easiest to integrate within existing company processes. For the pharmaceutical industry, that is computer vision technology. Computer vision technology has been consistently used for automated inspection toward the end of the production lifecycle. Our understanding and the maturity of this technology produce more value per unit risk than most other technologies that exist on the far right of the AI hype curve.
As we travel to the left of the hype curve, there is much more noise and enthusiasm, but the tangible use cases and actual value per unit risk goes down. I do see the use of the machine learning clustering model for data cleaning as an enabler for novel ways of working in the near future, as well as the use of generative AI to truly disrupt the regulated workforce in the next 3 years. Large language models and vector databases are wonderful tools to use for analyzing data from a human source.
Through this transformation, we cannot forget that people are still at the center of what we do every day to deliver therapies to patients.
Can you share a pivotal moment when you recognized the profound impact AI would have on your sector?
My first experience with AI was 14 years ago when a colleague of mine began building protein computer models. Back then, it took three months to generate 100 milliseconds of protein movement within a supercomputer. Analyzing how a protein moves in the body informs the potential efficacy of the therapy in use. Though our processing has already become much faster, that was when I first saw the promise of AI.
Natural language processing and large language models are not new, but in 2022, ChatGPT was released and became pivotal for the global democratization of AI. We have now put an AI tool in the hands of anyone who has access to the internet. What the general public can now create from ChatGPT has done more to push the drivers behind AI. I don’t think there’s any going back.
How are you preparing your workforce for the integration of AI, and what skills do you believe will be most valuable in an AI-enhanced future?
First and foremost, the answer is data acumen. The workforce of the future will need at least a basic-to-better understanding of what data is and how to structure and manipulate it.
In the past, the world’s political and social power source revolved around oil and who had control of oil. Today, the world’s power source revolves around data — those with the best data and the ability to analyze and withdraw answers. We are seeing many large technology companies halting the development of entire portfolios in order to focus on data and AI, which helps them see the impact on their future bottom line. From top to bottom, a basic to better data acumen is going to be a requirement.
What are the biggest challenges in upskilling your workforce for an AI-centric future?
In an AI-centric future, the biggest challenge to upskilling the workforce will be the loss of a large percentage of our talent pool. The baby boomer generation is larger than subsequent generations, and with that, there are more people leaving the workforce than coming into it. When that generation departs, they will take more than 40 years of accumulated knowledge that won’t be transferred to new talent at the necessary rate.
However, the knowledge vacuum following that generation’s departure is driving the adoption of AI faster. We have to create a lot of that experiential knowledge with AI, rather than waiting 15 years for it to happen naturally.
Finally, though basic knowledge is preferred, expecting every employee to be a program coder isn’t realistic.. Therefore, more no/low code or copilot AI applications will be needed to fully democratize the use of AI. While this will take some time, we’re seeing more and more technology companies releasing copilot tools with their products to boost adoption.
What ethical considerations does AI introduce into your industry, and how are you tackling these concerns?
Personal identifying information is a major ethical consideration. The more we use AI, the more data we require. As organizations start to scavenge for more data, data acquisition is going to get more and more intrusive. Efforts to protect personal identity and private information will be key.
Another key ethical consideration is the innate bias of AI models. To provide an example in pharmaceutical development, a company developed a computer vision algorithm for the detection of skin cancer, which requires patient images for inclusion. Nearly all the photographs submitted and used in the initial algorithm were from Caucasian patients. In later stages, the computer vision algorithm did not have the ability to pick up skin cancer in African Americans, Latino Americans, or other races. An AI control strategy must include a mechanism to root out biases.
Further, most of the global data produced is from more industrialized nations, and with that, the AI models will inevitably bias the answer away from developing countries because they’re not producing as much data. For example, answers and information used in ChatGPT are based on reading the internet. If you look at internet use around the world, the majority of usage is from the US and well-developed countries in Europe. As you go through your data set, you have to be cognizant of inherent biases.
What are your “Five Things You Need To Do, If AI Is Disrupting Your Industry”?
1. You need to pilot your own company’s efforts…yesterday.
There are a lot of people who say AI is coming after white-collar jobs, robotics are coming after blue-collar jobs, and soon, there will be no jobs left for humans. In reality, however, for every job lost due to AI technology, tens of new jobs are created. This was visible through the dot com boom.
Instead of fearing what is coming with AI, it’s better to think — how can I better position my company to survive? Whether it’s internal or external resources, you must begin piloting efforts to prove business improvement and subsequent return on investment.
2. Begin broad education within the company.
As I said, there’s a lot of fear and misunderstanding surrounding AI — having a base knowledge of data and AI throughout an organization is the only way to get rid of that skepticism.
When my company first announced we were going to tackle AI, especially generative AI, our team members came forward with concerns connected to existing conspiracy theories. When I held open forums, I was intentionally provocative, so I could later put together a plan to address those fears. I created an internal community around digital transformation and enablement in order to provide curated, accurate knowledge and a space for problem-solving and feedback.
Having an open and honest discord on AI and its impacts was important to our community.
3. Formalize a data strategy.
It’s interesting that this advice comes third and not first. At our company, there were a lot of people internally who thought strategy formalization should have happened earlier in the process. The reality is, at that point, we didn’t know what we didn’t know. The educational tools we used during the piloting program helped influence what our data strategy should be. As we began developing our data strategy and architecture, we already had a foundational understanding of the ethical pitfalls to guard against and the technical nuances we would experience.
From the beginning of implementation, I knew we couldn’t do this alone. I never once swung a golf club on a course before I could take a lesson because I didn’t want to screw it up. Partnering with our data strategy company was vital.
4. Have a very focused unit for AI.
It’s important to invest in a core team to root out inefficiencies within the organization that can be solved with better data. That team can then prioritize strategic data acquisition and analytics efforts.
As a company, you are producing data, but if you were to run analytics immediately, there would likely be data holes that create bias and intervene in providing the correct answer. To compensate for this, this team needs to think — what data sources do you need to bring in? What data or pipeline should you pay for to ensure accuracy?
5. Incorporate Data into job descriptions
By this point, your program has reached a level of maturity through the pilot program, proved value to the organization, and provided education tools to alleviate any fears, allowing for a widespread baseline understanding. You’ve established a data strategy foundation for upcoming work, implemented strategic data acquisition and formalized the data team that will bring all of this to fruition.
Now, the incorporation of AI will change the workforce as we know it. “Data acumen” becomes a skill you see in a job description — just like you see “strong communication skills” and “computer work.” Hiring teams will soon ask for this skill and challenge that in an interview. If data acumen is going to be embedded within the workforce of the future, we need to hire people with those skills today.
What are the most common misconceptions about AI within your industry, and how do you address them?
There are two sides to these misconceptions.
First, there’s the widespread fear surrounding the power of AI. How are we justifying its implementation? How are we showing the world that we can handle data and not have AI run amok? To address this, we need to confront those fears head-on with widespread education.
The other side of the spectrum is the people who use the term AI as a verb — who underestimate the complexity of AI. These are the people who say “Let’s just AI it”, thinking someone can somehow hit a switch, turn on AI, and it will solve a problem instantly. To alleviate that misconception, we need to put in an induction process for AI projects where we specifically look into the data inputs, understand the depth of the process, and go over the required quality benchmarks.
Can you please give us your favorite “Life Lesson Quote”? Do you have a story about how that was relevant in your life?
There’s a poem called “If” By Rudyard Kipling and the second stanza is
“If you can bear to hear the truth you’ve spoken; Twisted by knaves to make a trap for fools, Or watch the things you gave your life to, broken; And stoop and build ’em up with worn-out tools:”
I have never feared starting over in my life. The reality is you never start over from scratch — you start over from experience. My career has taken some odd twists and turns, and I’ve had bosses in the past who go, “Why are you making this decision to pivot? This will be terrible for your career.”
I would always tell them, “If it all fell apart tomorrow and I had to start all over again, I’m not afraid to make a decision to move forward because of failure.”
What thoughts or concerns often keep you awake at night? How do those thoughts influence your daily decision making process?
In all honesty, it’s typically my daughter’s well-being. But regarding this subject, there is something coming behind the AI boom — quantum computing. I’ve been spending a fair amount of time trying to understand this advancement. Several months ago, Google ran a simulation on quantum computing, and what took modern computers 47 years to run a computation — quantum computing did in six seconds.
Many believe we’re not limited by computation anymore, we’re limited by data. And that’s true. The introduction of AI to process existing data has significantly advanced research and development of novel therapies.
As we begin using quantum computing to develop new therapies, I fear the processing will be done at such an alarming rate that the supply chain can’t support it. The speed at which AI and quantum computing can support therapeutic development will lead to a massive stack-up of drugs that work and are ready to save millions of lives tomorrow, but we can’t get them approved, manufactured, and distributed in time to help those patients. It can be gut-wrenching to imagine a loved one desperately wanting to live and knowing there is a therapy that can work, but it’s still 5 years away from distribution.
If you could start a movement that would bring the most amount of good to the most amount of people, what would that be?
I’d start a movement to remove ego from the world. Ego suppresses free thought and decisions made by people and within organizations. Decisions are consistently being made for personal gain or in an “us versus them” gain. There’s a lot of bottlenecking in organizations due to corporate pride.
For example, I have heard many company leaders say, “AI will never replace me or I will be replaced.” That self-centered, egotistical approach will cause us to make very narrow, short-sighted decisions that ultimately do more harm than good.
How can our readers further follow you online?
Thank you for the time you spent sharing these fantastic insights. We wish you only continued success in your great work!