Real Technologists Newsletter

Thursday, February 22nd, 2024

Thursday, February 22nd, 2024

John Kaufhold

ML Journeys are all different, but they rhyme: Wherever you are on your business’s ML journey, Dr. Kaufhold has been there in his 30 years of ML experience. Dr. Kaufhold first mastered his multidisciplinary approach to ML in his roles as independent contributor and research and development leader at General Electric’s Global Research Center, SAIC (now Leidos), and NIH. In those roles, Dr. Kaufhold saw it all — whether it was blood vessels in mouse brains with neuroscientists at University of California, San Diego (UCSD), or wiring diagrams of brains on sabbatical in computational neuroscience at MIT, or cargo inspection with Department of Homeland Security’s Global Nuclear Detection Architecture program, or mammographic breast density estimation with the National Naval Medical Center, or aligning diverse OWL2 databases for the intelligence community with IARPA, or target recognition of relocatable ground vehicles in radar imagery for DARPA, or identifying a species from a photo (on a phone in airplane mode) for DARPA, Dr. Kaufhold’s been leading teams small and large to successfully prove ML and generate ML ROI for three decades, most recently for the startup he founded in 2013 and later led through acquisition, Deep Learning Analytics.

Episode Transcription:

Trac Bannon: John Kaufhold is one of those super smart people… you know the ones… when you read their LinkedIn profile and do a little googling, their achievements and academic records are phenomenal. I was initially a bit shy about reaching out to talk with him. 

I’m a techie. I’m not a data scientist though I’d like to think that I’m math and AI adjacent. John’s contact information was shared with me by Jennifer Ives, CEO at Watering Hole AI and a Real Technologists alumni. 

When I hopped on the call with John, he surprised me by being a few minutes early in hopes of meeting Bob. That’s become a theme with guests… on this particular Friday evening though, it was only John and me. I was immediately drawn to John’s personality. During the rest of our chat, I figured out why. Like me, John grew up in what we like to call “the rest of Pennsylvania”, the part that is not metro Philly. 

John grew up in Lancaster, Pennsylvania. In fact, his wife Liz, who he affectionately calls “the boss” grew up there as well. Lancaster is steeped in U. S. history and is known for its well preserved historic buildings. Lancaster is also known for its large Amish population. It has one of the oldest plan people communities in America. 

John describes his upbringing as typical blue collar. His parents loved him deeply and placed an emphasis on education sacrificing to send him to a private Catholic school. 

John Kaufhold: Second street. Yep. And the boss, she grew up in Schooling Hills, which is like the, you know, the nice part of town. I went to Sacred Heart school. My mom and dad. my mom was a Bell Telephone switchboard operator. And my dad, he was a radar operator in the Navy. And then he got a job as a network technician. And neither of them had a college degree. So it was kind of a big deal for them to spend that kind of money to put Mike and I in Catholic school. 

Trac Bannon: His parents, especially his mother, invested not only the funds for John to attend Sacred Heart, she invested a ton of time. She made sure John and his older brother Mike were set on a solid educational path. She did all this while battling illness. When John was still in elementary school, his mother was diagnosed with breast cancer. 

John Kaufhold: What my mom did as she was dealing with 10 years of… recovering from mastectomy and chemo and just the slow process of breast cancer. She was making sure that Mike and I, we were her project… and she made sure that we got all the educational opportunities she could possibly give us. So she put us in these programs at the local college, Franklin Marshall… and we would take… like some kind of Saturday classes and it would be us, these two very blue collar kids, and then all of the rich kids in class. 

Trac Bannon: You are listening to Real Technologists. I’m your host, Trac Bannon, coming to you from Camp Hill, Pennsylvania. Each week we choose a unique guest behind leading Edge Tech innovation to explore their genuine stories, their true journeys. Technology touches nearly every aspect of our lives. It’s being driven by diverse perspectives and experiences of real humans.

You’re in the right spot to hear about the real technologists reshaping our world. Stay tuned for stories that will give you something to noodle on.

John was affected by his mother’s journey and his father’s character. You can see their fingerprints in John’s character and you can trace their actions to his own journey. Those college programs, coupled with John’s natural love of math, set him down the STEM path.

John Kaufhold: Dad was awesome. Dad, he was kind of the person who wrote the blueprint for our character. He would always encourage Mike and I if we were having trouble with anyone to consider what they were going through, try to see it from their point of view. I think those were really good. you know, parents, for me to have.

Trac Bannon: I’ve gotta wonder if it was his Pennsylvania blue collar background that gave him so much spunk and determination. Somewhere around 9th grade, he started playing volleyball with friends. At 5 foot 8 inches tall, he was always the shortest person on the team… so what? He loved the game. And he played in rec leagues, JV and varsity at McCaskey High School, and even played on the Junior Olympic team. He split his free time between volleyball and stocking shelves at the Weis Market. I had to laugh when he mentioned stocking shelves. My brother, the building architect, stocked shelves at our small town grocery store as well… the Super Duper. 

John’s mother went to bat for her boys. When she saw biases against them, she moved them to McKaskey Public School. During his junior year, John’s mom enrolled him in a summer school program at the University of Mass Lowell. She knew he loved math and would like engineering. 

John Kaufhold: A few weeks before it started, they just phoned her and said… sorry, we got no housing for your son. Uh, can’t come here. So this whole thing that she had orchestrated for me wasn’t gonna happen. So she said to them.. Well, then you’ll find somewhere else for him to go . And so she, uh, she had them… you know, call around, and BU was the program that I ended up going to. I spent that summer in my junior year there, and I was like, I’m going to go to school in Boston. Boston’s where all the kids are.

Trac Bannon: John is very open and honest, the type of guy who readily admits and owns his decisions. Like many young teens, John felt estranged from his dad. The stress and strain of his mother’s ongoing cancer battle must have been an immense strain on the entire family. That said, John stopped talking to his dad mid teens. 

John Kaufhold: In my senior year, my mom passes in September and I had not been speaking to my dad since I was 13. And my brother had just moved away to college the prior year, so I was kind of on my own, and then I was in such, I don’t know… antagonistic… I don’t know… difficult teenager. What do you call it?

 January, my dad went away to some kind of training for work. And I just asked a friend if they put me up for a while. And so I moved out when he was away, came back, he wasn’t there. Probably one of the worst things I ever did.

Uh, can you imagine if I was in pain? Can you imagine the pain he was in? He just, I mean, like… you can look at the statistics on what’s stressful in a person’s life, losing your significant other. That’s what just happened to him. I lost my mom, not, not nothing, but, and then for the kid to just…. disappear when you’re away. 

Trac Bannon: There have been many influential people in John’s life. Interestingly, the majority of the ones John told me about were and are women. His mother’s life and death left an indelible mark. So did Becky. Remember that Boston University summer school? Becky was one of the college students that coordinated the program. When John arrived at Boston University, Becky opened doors and made opportunities. 

They were both biomedical engineering students and shared a very personal life event: both had lost their mothers. Becky would look out for John, helping him to land a job moving furniture for university housing and introducing him to the boss at the lab where she worked. John even became a program coordinator for that high school summer program that he had attended. You could call Becky a sponsor… someone making connections and pointing out resources.

John Kaufhold: So I got a job there. And that’s when I started in speech recognition machine learning.

So if there was one person that pointed me at my machine learning career. It was… I can’t give her credit for this being intentional, but I did happen to do machine learning for the rest of my life. Becky also had lost her mom, like I did at about high school… went through the BU BioMed program, did her senior project in that same lab. So a lot of, a lot of similarities and I still keep in touch with Becky. She’s a professor at University of Minnesota, Mankato.

Trac Bannon: John went on to become a prestigious Witaker fellow having been nominated by Laura Carney and Ken Lucien… both well known to the people in the biomedical engineering community. 

Imagine his career excitement not to mention the glee of going from $7.50 an hour moving furniture to making $16,000 a year to research. The millennium change came, and after nearly nine years, John graduated with a PhD in biomedical engineering. 

John’s first job out of college was with GE as a biomedical engineer. His focus? Image processing and segmentation. 

John Kaufhold: I’m at GE… that first year, I’m supposed to go to RSNA, the big radiology conference in Chicago. And at that conference, I’m supposed to network with radiologists and figure out how we’re going to do collaborations and be successful at whatever I’m doing at GE… which is mammography, by the way. And the whole impetus for me saying, Oh yeah, that job, I really want to do that because I want to make it make a contribution in mammography. That’s what my mom passed from… so that was a great, opportunity to say- this is what I’m going to do. So, that let me kind of check a box and kind of free myself a little bit of like, I did something there. 

Trac Bannon: John does not mince words when talking about his wife, Liz. You can hear with each word he says about “the boss” that he truly adores her. It’s not often these days to hear about a couple that grew up in the same small town, attended the same church, and remember the same kids and teachers from high school. As fate would have it, Liz attended BU as well. One day he saw Liz standing at BU Central, at the e-stop, Liz’s cousin, Jay Apple, had been in the high school math club with John. He seized the moment. 

John Kaufhold: Hey, aren’t you Jay Apple’s cousin? she was way out of my league. Like her dad was the head partner at the law firm downtown, Apple and Yoast. And he went to Harvard Law School with Anton and Scalia. I had a mullet in high school and my parents did not have even a bachelor’s degree, right? So I am on, I’m the wrong side of the track. 

Trac Bannon: Liz and John would occasionally run into each other and exchange emails. Liz went off to law school and their relationship plodded along through those occasional emails. That is, until John got his first job with GE. He was tasked with attending a big radiology conference in Chicago. Part of his responsibility was supposed to be networking, or as he calls it “Schmoozing”. He felt ready to give up and went for a walk. 

John Kaufhold: I just go for a walkabout… I’m walking down Michigan Avenue, I look up and I say, Liz? And she says, John? What are you doing here? And, thinking to myself, like, uh, what are, what are the chances that this person like… if you took the XYZT coordinates of our lives, there’s no person on the planet with a bigger dot product than her, and there she is. I just randomly run into it. That’s like fate hitting me over the head with a frying pan thing. You’re really supposed to talk to her. That’s like the best thing that ever happened to me. And again, that was one of the things that, it started as me in a moment of professional failure. Not knowing who to network with in Chicago, right? 

Trac Bannon: They began dating, though they decided that they would pull back on building a relationship since she was practicing law in West Virginia and he was in Albany, New York.

Then John took a chance. He was headed to Germany to give a talk about mammography and gave her a call. Long story short, she agreed. She waited until she joined him in Germany to surprise him. She got a job in DC, and there are direct flights from Albany to DC. They were married in 2010 at Lamont Farms in York, Pennsylvania.

There’s a certain trait with many Real Technologists that inspires me . They take a chance. They say yes to opportunities. Their journeys are dramatically changed because they took a risk. 

 John stayed with GE for 5 years before moving to SAIC, the Science Applications International Corporation. SAIC is a Fortune 500 company that primarily serves the U.S. federal government, including the Department of Defense, Department of Homeland Security, and other federal civilian agencies. John would spend nearly 8 years as an image scientist and technical fellow. 

During his tenure at SAIC, John led and contributed to a range of innovative projects that combined machine learning, image analysis, and data fusion. How about enhancing security by analyzing cargo container images? Or advancing medical research through 3D brain imaging? John even developed algorithms for camouflage patterns. 

His list of accomplishments is long. Remember, I said this is one of those uber smart people that I need to get a little courage to call? John took a sabbatical from SAIC to study Deep Learning at MIT. This was 2010, well before what John calls the Deep Learning Big Bang. 

Next, it was time to puddle hop to the National Institutes of Health. The role and the leadership were just not as described. 

John Kaufhold: I remember in my exit interview, I was talking to the woman that actually interviewed me when I came in. She’s like, so where are you going from here? I’m like, I don’t know. Does that make sense? She’s like, no, it does not make sense… like I have no plan. I’m just, I’m just quitting. And I said, what if I started a company? Does that make sense? No, also does not make sense. But that’s what I ended up doing.

Trac Bannon: It was his wife Liz that pushed John to start his own company. She knew it would consume his time. She knew it was a risk. 

John Kaufhold: Use that network. Start your company here. If you could take your company to Lancaster, if that works out… you could do all these other things. But she’s like, if you don’t do it, you might regret it. Oh my God, she was right.

And she also made it clear that I had every license to fail… and fail spectacularly. She gave me the ticket to the best professional experience in my career, had faith in me, more faith than I had in myself.

Trac Bannon: John founded Deep Learning Analytics, DLA. It wasn’t an easy slam dunk. Working with the U. S. government is fraught with policies and contracting rules so complex that sometimes even the government officials are not clear. To fund the new startup, John kept working on side deep learning analytics projects. John sought advice from a Score mentor named Gerry Sanz. Score is a non profit group partnered with the Small Business Association, the SBA, to help small businesses. John’s wife suggested he work with Sue Rego. She had run companies before and would balance John’s technical focus. 

John reached out to a government program manager to have a conversation. The topic, john’s ideas on how deep learning techniques could improve radar images. That bold move landed John with a feasibility study. 

The study drove the government to release a broad announcement asking other companies to evaluate and expand on the original study. His study had created a market that seemed beyond his own reach. Major players were lining up. John decided to put in a proposal… this was his idea. It was a long shot. Nearly everybody that advised him said it would be impossible to submit and win. Apparently, John and Sue were up for the challenge. 

The story of choices they made could make a made for TV movie . Imagine proposing experts that you had not yet hired and winning against the likes of Lockheed Martin, Boeing, and Raytheon. Once the award happened, it was time to hire. 

John Kaufhold: These were people I’d worked with before. I knew them. And this is one of the, I guess, bits of wisdom in my career is be a good teammate. Because these people were the most amazing ML engineers and data scientists and software engineers that I’ve met in my career. I thought, because it was such a risky prospect, a company with no past performance doing, you know, can’t even… like, we weren’t expected to submit a compliant proposal, let alone win… let alone win phase two after winning phase one. So it was all risky. It’s a startup. These people had really good jobs at stable companies. Every single one of the most amazing people joined.

Trac Bannon: John’s approach to hiring was innovative and intense. He reached out to the best candidates, the best thinkers, and gave them a puzzle to solve. He handed them an NVIDIA Jetson board designed for embedded AI systems and IoT, Internet of Things. His instructions? Train and test a model that fits on that small profile hardware. 

John Kaufhold: The whole interview project was take this piece of hardware, put a model on it that was trained on some data. And then test it on data… evaluate its performance on data that it wasn’t trained on. Write it up, send it back, and we’ll tell you whether you’re in or not. And it was very clear who wanted to be in. That was a novel way of interviewing, but it reflected what we were actually going to do in the project. And it was also confidence building, because anybody at that time might have seen deep learning as this hocus pocus, crazy, mysterious thing that was a big black box.

Trac Bannon: Within six months, the DLA team was delivering. He still calls some of the timing luck, however. In 2012, no one was focused on hiring the type of data scientists that he needed. John is openly proud of the team he built and the methods he used to interview them. One of the biggest sources of pride is having sourced the candidates from a diverse pool. John is an authentic advocate and ally for diversity, regardless as to the bend. Age, education, gender, race, or self identification. He refers to the cognitive advantage of having a multi shade team. 

John Kaufhold: Things I’m proud of are that it was half women, half men… that’s something that was really unusual. We had a 0 percent attrition rate for 7 years. No one left, everybody stayed. Everybody said, I’m in this… this is a good thing. And I give credit to the rest of the team for that because most of their interactions were with other people in the team, not me.

Trac Bannon: The company grew to one of the largest deep learning teams in the country before John ultimately sold. Why sell a hot deep learning company that was even featured in the Washington Post? The answer is that John cared deeply for his team and had a keen understanding of their incoming work pipeline. In the same way he was so proud of their accomplishments and lack of attrition, he was also aware that time was marching forward for everyone, adding layers of complexity.

John Kaufhold: Obviously it’s in my blood to be a scientist and it’s what I was trained to do. But once you have a startup, you are focused on just the health of the startup and the employees. So when everybody starts DLA, nobody has children, nobody has a mortgage, everybody’s free. By the time, we’re shopping the company around for offers. People are married, have kids, and every one of these is a burden. I had to bring in the money. And when I saw the amount of money shrinking that runway… well, that’s just getting more and more stressful for me. 

Trac Bannon: John met with M& A advisors and soon had a phenomenal offer from General Dynamics. They wanted the first day at GD to be just like the last day at DLA. And they absolutely made good on that. The DLA team even kept its name. They became the General Dynamics Mission Systems Deep Learning Analytics Center of Excellence.

John stayed on with GD for two years. And during those years, he continued to mentor and nurture the team all the while, turning over the reins to others. 

John is very patient and indulgent with others, though I could tell there is both a mix of frustration and excitement with the rapid acceleration of AI. John is concerned about the pop up experts, those who toss around buzzwords and have not spent time to truly learn, experiment, and discuss with AI experts. John is deeply concerned about the focus on Generative AI and the creation of new content. And he’s also concerned about machine learning algorithms. ML algorithms make suggested decisions, for example, who should get a loan? 

John Kaufhold: Now empathize with me. I’ve been doing this 30 years and, everybody has just discovered AI and become an expert via YouTube, in the past six months. So like, it was difficult before… now it’s near impossible to actually talk about ML use cases, how to monitor performance, how to vet them… how to make sure they’re safe, helpful, useful, that they don’t spill data in ways you don’t want them to. It’s gotten so much worse.

Trac Bannon: I feel professionally and personally blessed to meet this Pennsylvania native from just down the road who happens to be a data science legend. With all his book learning, research, and amazing discoveries, I was still able to catch an additional glimpse of John peeking through when we talked about our lives.

And advice we may give to our younger selves. He simply said that he wouldn’t have pursued academia so much. In his words, “it just ended up being that they didn’t want me.” 

What is on the horizon for this data scientist with the signature flyaway hair? After leaving General Dynamics in 2021, he and Liz spent a year or so building a new house.

John Kaufhold: We built a house and, that was a good deal of work. We moved in basically the beginning of 2022. So that was its own effort of just like constantly talking to the builder and all the subs about that. And then there was a year of basically once you build a new house, you do all the things that make it actually livable.

We love this house, but I’m now on my fourth list of house projects, and there are thousands and thousands of lines of things… just Little tiny things. I am definitely an engineer. Like I do build things and I do Make sure function and form fit, like basically the whole house… she made every decision on how something looks and thank God, I’m like, well, can we get the roof to be 40 degrees to maximize the solar production? Uh, and face due south. 

Trac Bannon: John admits he is always a scientist; it’s part of his DNA. He has not walked away from data science, quite the opposite. He has strong words for all of us… thank heavens he’s back at the game, advising folks.

John Kaufhold: But ChatGPT is historic. Just the uptake, the way that penetrated 100 million users inside of a month is bigger than any other phenomenon we’ve ever seen… doesn’t mean people know how to use it. If you’ve discovered ML in the past few months, you’ve discovered just the UX of ML, not ML. Like, it used to be that people tried to hide ML from the user. Now you get to interact with it directly. And people are scared because they didn’t realize that there was that power, that predictive power behind the tools they were using before, but it is really just a tool.

 So, the biggest thing is going to be our social fabric… how do we respond socially to this technology ? Sounds boring, but it is, and probably the most trying time socially for the adoption of the technology. I have started consulting again, helping companies understand what they need to do to get machine learning into their products and services.

Trac Bannon: And that’s a wrap for today’s episode of Real Technologists. I want to thank my guest, John Kaufhold, for sharing his story. Your insights and experiences are truly inspiring. I’m grateful for the opportunity to share them with the audience. This podcast is a Sourced Network production and updates are available weekly on your favorite audio streaming platform. Just search for real technologists and consider subscribing. Special thanks to our executive producer, Mark Miller, for making this show possible. Our editor and sound engineer, Pokie Huang has done an amazing job bringing this story to life. Thank you both. The music for today’s episode was provided by Blue Dot Sessions, and we use Descript for spoken text editing and audacity for the soundscaping. The show distribution platform is provided by CaptivateFM making it easy for our listeners to find and enjoy the show. 

That’s all for today, folks. This is Trac Bannon. Don’t forget to tune in next week for another intriguing episode of Real Technologists and something new to noodle on.

Episode Guest:

John Kaufhold

ML Journeys are all different, but they rhyme: Wherever you are on your business’s ML journey, Dr. Kaufhold has been there in his 30 years of ML experience. Dr. Kaufhold first mastered his multidisciplinary approach to ML in his roles as independent contributor and research and development leader at General Electric’s Global Research Center, SAIC (now Leidos), and NIH. In those roles, Dr. Kaufhold saw it all — whether it was blood vessels in mouse brains with neuroscientists at University of California, San Diego (UCSD), or wiring diagrams of brains on sabbatical in computational neuroscience at MIT, or cargo inspection with Department of Homeland Security’s Global Nuclear Detection Architecture program, or mammographic breast density estimation with the National Naval Medical Center, or aligning diverse OWL2 databases for the intelligence community with IARPA, or target recognition of relocatable ground vehicles in radar imagery for DARPA, or identifying a species from a photo (on a phone in airplane mode) for DARPA, Dr. Kaufhold’s been leading teams small and large to successfully prove ML and generate ML ROI for three decades, most recently for the startup he founded in 2013 and later led through acquisition, Deep Learning Analytics.