When lives are on the line, building the right infrastructure changes everything. Dr. Serena Huang explores how AI is transforming cancer research and patient care with Jessica Audette, Head of Enterprise Infrastructure Strategy and Transformation at Memorial Sloan Kettering Cancer Center.
When the stakes are as high as cancer care, the infrastructure powering the research is becoming as critical as the science itself. Dr. Serena Huang sits down with Jessica Audette, Head of Enterprise Infrastructure Strategy and Transformation at Memorial Sloan Kettering Cancer Center, to examine how MSK's supercomputer, Iris, is compressing years of research into months. They discuss how AI is transforming patient scheduling, clinical documentation, and at-home recovery, and what organizations in regulated industries need to consider before deploying agentic AI.
Timestamps:
1:39 - Meet Jessica Audette
5:28 - Iris: MSK's supercomputer ranked #2 in the US
8:45 - AI-powered scheduling for patients
11:47 - AI dictation for clinicians
15:00 - Personalized medicine through wearables
17:35 - Meet Amelia: MSK's agentic help desk agent
20:55 - Safety, compliance, and working with InfoSec
22:47 - Strategy before deploying AI
29:58 - Rethinking what innovation really means
Links:
[00:00:00] Jessica Audette: Most folks are mystified by technology and the capabilities it can offer. We often have this Veruca Salt mentality of, if you've seen Willy Wonka, "Don't care how, I want it now." The researchers don't want to and shouldn't need to figure out how to get from point A to point B. We want them to focus their efforts on ending cancer for life.
[00:00:30] Serena Huang: Welcome to Deep Geeks. I'm Dr. Serena Huang. We spend a lot of time in this industry talking about speed, faster inference, faster training, faster iteration But there's a place where speed isn't just about competitive advantage. It's about whether a patient gets a diagnosis in time. My grandma survived cancer twice.
She is 94, and I just got back from visiting her last week, and I've been thinking a lot about her in preparing for today's conversation because what my guest is building is for people exactly like her. Our guest today leads one of the most ambitious research infrastructure teams in the country at Memorial Sloan Kettering Cancer Center, and she is here to talk about what it actually takes to build at that scale, what's still in the way, and what AI looks like when the stakes are as high as they get.
Jessica Audette, Head of Enterprise Infrastructure Strategy and Transformation at Memorial Sloan Kettering. Now, as I look at your background, infrastructure and cancer research, that's not a typical combination I see. So I'm curious if you can share with us how you got here, and perhaps more importantly, what kept you here.
[00:01:58] Jessica Audette: Well, thank you for having me, Dr. Huang. I am really thrilled to be able to share, uh, about what-- the work we're doing, and I certainly did not have a linear career path. Like many millennials, I graduated at the height of the recession. I remember my first job, I was making $10 an hour in Brooklyn.
[00:02:19] Serena Huang: Wow.
[00:02:20] Jessica Audette: So back then, I used to know all the places to get free food at.
But technology has really always been a passion of mine. I think, again, being a millennial, we were the-- on the cusp of zero tech to the days where your mom would be telling you to be sure that you're home when the streetlights go on, to the tech boom with PCs and the World Wide Web. So that's kind of where my love started.
Yeah. Largely, my career has been in the financial industry, yet a colleague reached out to me about an opportunity at MSK, and this was just a thrill for my parents, who have been in the medical field as physician assistants since 1985. Wow. And at least one of their three children would be in the medical field, albeit a degree of separation, but, uh, they were just tickled by that.
But as you said, I think that there aren't many of us that haven't experienced cancer or been directly affected by cancer or know someone that has. So the mission of ending cancer for life, MSK's mission, is a noble mission to stand behind, and that's what ultimately keeps everyone coming back.
[00:03:31] Serena Huang: I love it.
Yes. I think we need, we need that mission everywhere. Um, I can certainly relate to it. I know our audience can as well. In our preparation, you said something very provocative, at least for me, as a recovering data scientist. You said algorithm is not the real bottleneck in cancer research. How can algorithm not be the real bottleneck in cancer research?
Can you tell us more, and what is more important than that?
[00:04:02] Jessica Audette: It's true. Algorithm is not the bottleneck. Infrastructure is. Research has really been going on for centuries, right? But the differentiating factor between the researchers of today and yesteryear is that today they have cutting-edge, high-performant infrastructure platforms to help them advance their science and their research to really aid them in solving complex problems around image analysis, signal processing, complex modeling, and data analysis.
So I think when we say algorithm is not the real bottleneck, we can work with the algorithm. We can work to enhance the code and, and the workflows and make sure that all happens, but without the platforms, there's not really a place to put those things.
[00:04:47] Serena Huang: I think a lot of us forget that in particularly cancer research, there is actually a large amount of data.
If I think about images, right? This is not a small data set with a few lines of data, right? This is not just a small spreadsheet we're talking about, and I can only imagine, uh, as a researcher, how much of that really takes for discovery purposes and to be able to eventually diagnose, is this cancer or is this something that looks like cancer but isn't cancer?
Well, now that we know why algorithm is not the real bottleneck and infrastructure is, I want to pivot to talk a little bit more about the specifics. Iris, the supercomputer which is ranked number two in the US and number four globally. Congratulations, by the way.
[00:05:43] Jessica Audette: Thank you.
[00:05:44] Serena Huang: And you have told us that it's able to reduce wall clock time by thirty times.
Thirty times. So what would have taken years now take only months in discovery We have some audience who maybe never thought about supercomputing before today. What does that mean? Can you walk us through what it means for a researcher who's sitting in the office on a regular Monday? Um, maybe in the process, did anyone underestimate how important and the power of Iris?
[00:06:17] Jessica Audette: Absolutely. So let's start at the beginning. When we were designing Iris, we decided to overhaul our entire technology ecosystem, moving away from legacy platforms and tools, consolidating what made sense, and implementing the best-in-class solutions for life sciences and AI ML workloads. Ultimately, we wanted to maximize performance and the amount of research Iris could support.
So what does that mean for the researchers, uh, that are in the labs and such? Well, most folks, myself included, at times are mystified by technology and the capabilities it can offer. We often have this Veruca Salt mentality of, if you've seen Willy Wonka, don't care how, I want it now. The researchers don't want to and shouldn't need to figure out how to get from point A to point B So we'll put it this way.
If everyone wants to own a Ferrari, if there's no road to drive it on, you're never gonna get really far, right? We don't want the researchers to waste their time thinking about that infrastructure and how they're gonna get from point A to point B. They want that road there yesterday, and we want them to focus their efforts on ending cancer for life.
So ultimately, we've deployed a cutting-edge network featuring high speed and low latency, as well as several different generations of high-performance computing nodes. But the major change that we implemented was our primary file system. We implemented WEKA, which is a modern enterprise-class file system.
It runs on NVMe-based flash storage. Uh, it has cloud-bursting capabilities for some of the users that are interested in that. But ultimately, the architectural choices have enhanced our data processing capabilities, streamlined our research methodologies, and provided a high-class platform for groundbreaking discoveries.
So essentially, we've built the autobahn for the Ferraris to drive on.
[00:08:19] Serena Huang: I love it. That's a perfect analogy. In-indeed, if there's no highway, then there's no place for your Ferrari to go. Um-
[00:08:28] Jessica Audette: Yeah, but just park it in the garage.
[00:08:30] Serena Huang: Yes, exactly. It would just sit there, beautiful, but not, not helpful. Let's go to some use cases now, because most importantly, what matters is how does this make a difference in the patient's life, in the cl-clinician's life?
And I know you mentioned scheduling as one of the use cases, and I can't tell you how many times I've had to reschedule and call different offices, uh, for something simple. And I can just imagine, especially for patients who are, um, wondering, "Do I have cancer?" Or, "Now that I know I do, oh my gosh, that's a huge news.
How do I get more scans, get treatment, and schedule things?" The last thing you want to figure out is, how do I get to the doctor's office? Can I get to the doctor's office on time? And oh my God, I have to take time off, and can I make it there? So can you walk us through a little bit about what scheduling, um, is like powered by AI?
[00:09:35] Jessica Audette: Sure, happy to. So in any medical organization, scheduling is kind of the entry point for most. When undergoing cancer care, you often have a team of clinicians that work cohesively on your case- Hmm ... whether they are requiring imaging, testing, diagnostics, things of that nature. We have several locations in and around New York City.
[00:09:56] Serena Huang: Hmm.
[00:09:56] Jessica Audette: We've implemented predictive analytic AI for scheduling. So this really optimizes the patient and clinician's time. What do I mean by that? Well, if we've all been to New York, and you know that our main campus is located at 1275 York Avenue, right? We also have a building on 55th, and then there's, uh, several others within the area.
But if you're doing imaging at main campus and then need to meet with a physician at 55th, it's not gonna take 10 minutes to get downtown. We know that it takes longer than that by cab or train with, with traffic and such. So those types of predictive analytics afford the schedul- schedulers the opportunity to really optimize the timing so that we don't have missed appointments, we aren't overlapping or making expectations of the patient that they can't be somewhere or in two places at once.
So it really, really allows the clinicians to spend the time where it's needed most, which is at the bedside of the patient, as opposed to missing those opportunities.
[00:11:01] Serena Huang: I love that. Um, yes. I can think about how I used to have to, you know, use Google Maps on top of, you know, figuring out calendar just to see if I can get through traffic and get somewhere.
So that's amazing. Um, and I didn't think of the time that clinicians are waiting too, 'cause that's also time wasted before. If they wait 'cause someone didn't show up on time or someone is late, and then now everyone after that patient is late and have to wait longer.
[00:11:32] Jessica Audette: It's a domino effect- Yes ... for sure.
[00:11:34] Serena Huang: Yeah.
[00:11:34] Jessica Audette: And then, you know, there are other things that we're looking to alleviate administrative burdens on our clinicians as well, like dictation.
[00:11:43] Serena Huang: Hmm.
[00:11:43] Jessica Audette: So we've implemented AI voice recognition dictation to aid them in creating Really well-rounded documentation for the patients. So you go in, when you depart from meeting with them, they'll dictate all of the meeting notes and-- or the case notes, and then that goes to your portal that you're able- Wow
to spend your 20 minutes reviewing later on.
[00:12:06] Serena Huang: That's amazing. Is that something like where patients will also have to consent to being recorded, or is it just on the clinician's side that they are, um, recording the diagnosis or the, uh, the visit?
[00:12:21] Jessica Audette: Yeah. So the patients don't have any touchpoint with any of the dictation.
That's an internal process. So if you think back to, and my parents are guilty of it, they had medical charts. You'd see these huge filing cabinets with your last name somehow pinned onto it, and they would physically write out their charts. I remember them being home at dinner writing out charts- Oh, wow
from their day, and that would get filed away. This really, uh, optimizes the clinician time because as soon as they walk out of the room, they're able to sit down, dictate their notes, uh, and then move on to their next case, as opposed to waiting till the end of the day to do all of their case notes. And some of them may do that, you know, still, and do their case notes at the end of the day and such.
They all have their own routines. But that particular AI platform affords them the opportunity to say what they need to and move on. Right?
[00:13:12] Serena Huang: Between the scheduling and then being able to use, um, the voice notes at the end of a visit, have you noticed any changes in the patient experience? I'm all about data, of course, so I'm curious if there are any metrics that you can share with us that has changed as a result of this AI.
[00:13:31] Jessica Audette: What I will say is that we've obviously gotten resounding feedback from the patients about the, the scheduling options that are available to them. I will share that my mom was a patient of MSK, and she had a fleet of d- clinicians working on her case, and she actually received her treatment in Monmouth County in New Jersey.
Often, she would be in New York City and then need to be out in New Jersey for, uh, immunotherapy or vice versa. I would say, as a patient caregiver at that time, having those things available where I wasn't needing to call and tell them, "Hey, I can't make it to Monmouth County by 1:30. We're in Manhattan still, and she's still getting her immunotherapy or her radiation, and we're going to miss this appointment," that was all kind of handled for me.
So in that use case with my mom and, and my dad and I being patient caregivers, that was a relief because- Right ... I was able to focus on keeping my mom's spirits up and spending time with her and just enjoying- Yeah ... those moments as opposed to trying to deal with all of the scheduling madness.
[00:14:39] Serena Huang: Absolutely.
That is so relatable. Thank you. Thank you for sharing. So we talked a little bit about another type of AI, so personalized medicine, because wearables is all the rage, and many of us have AI that tell us how we're doing today and how we can optimize our health, depending on which spectrum of health we want to optimize, but there are lots of options now.
And I'm curious if you can talk a little bit about personalized medicine and wearables, and from where you are sitting, how is it making a difference for your patients?
[00:15:18] Jessica Audette: Wearables have been on the scene for a little while now, but they have really ramped up in adoption. I've seen several different, uh- options coming out to the market now.
We've implemented wearables at MSK a couple years back now. I believe it was 2022 or 2023. But this really changed the patient experience. So what our wearables do, and mind you, it's not for everything, it's not, you know, like a specific watch that does all sorts of things, but this is purely for the clinical, uh, needs of MSK.
We've implemented these wearables, and it's changed the experience for our patients because it's able to provide the clinical team with real-time data while recovering from the comfort of their home, and so we call it care at home. Um, cancer treatment inevitably causes patients to become immunocompromised, and in certain cases, it may be best for them to recover outside of the hospital.
So this program allows them to do that, be with their family, be in the comfort of their homes, and be able to still transmit data back to their clinicians and their care team. The clinicians at minimum are getting baseline, uh, vitals. So that's helpful in the recovery process is that the clinicians can track that in live time to ensure that they are recovering in the best possible manner.
Care at home affords these patients the opportunity to recover in the comfort of their homes with their family surrounding them, and it also keeps those folks that may be more immunocompromised out of the hospitals
[00:16:57] Serena Huang: Okay. That is important because, um, you don't want to risk their recovery if they are not-
[00:17:05] Jessica Audette: Exactly
[00:17:05] Serena Huang: they're better off at home. That's perfect 'cause I-- we, we don't think about that. Sometimes we think hospital is the best place for care, and for certain patients, that simply is not the case. Shifting towards another super hot topic, agentic AI. Um, it is everywhere now. It seems like everyone has agentic AI in their company, in their home, and I'm curious about agentic AI in research.
We talked a little bit about Amelia in our prep call. Could you share with us what does Amelia do, and how is it making an impact at MSK?
[00:17:43] Jessica Audette: In the digitized world, we're all expecting really fast responses, and that's no different in our organization where our customers w- like, whe- um, whether it be the physicians or other technologists and such, we want fast responses to our tech-related issues, so, uh, so much so that it's faster than humans can deliver when dealing with thousands of tickets across an organization as large as ours.
Uh, in this case, we deployed agentic AI named Amelia. This is our internal agent that works on our help desk. So we wanted to reduce call wait times and redundant work. Essentially, our human resources were focusing on a lot of what we call level one issues. I lo- got locked out of my computer, I need my password reset, can I get an account for this?
Can you get me access to that? By implementing Amelia for our help desk, we reduced our call wait time from forty-two minutes down to a minute with about a thirty-five percent cost savings for- Wow ... the organization. What does that mean? Because a lot of people are like, "Oh my gosh, my job's... AI is going to take my job."
But this opportunity allowed us to elevate human resources to focus on high-level solutioning. So Amelia will handle those redundant, "Hey, I got locked out of my computer," or, "This printer isn't working." You know, "My touch... tap to go is not working. Can you send an agent?" It'll either open a ticket for a higher level agent to come out and service, or it'll handle it itself, so reducing the call times and that ROI for the investment.
Now, one thing, again, where fear, fear-mongering comes in is like, okay, where are the graduates going to go now that, you know, L one used to be the entry level for graduate students, and that it's going to become more difficult for them to break into the workforce. They're-- I don't think that they're that wrong when it comes to that.
I think there's going to need to be a fundamental shift on the education side of things, but I think that this is also going to be a transformation in the way we work. So what would've been L-1 and entry previously will now be maybe it's L-2, and that's the entry point for our graduate students after they depart from school.
[00:20:08] Serena Huang: Yeah, absolutely. Wait, and did you say 42 minutes to one minute?
[00:20:15] Jessica Audette: 42 minutes down to about a minute Wow Yes, with 35% cost savings. Yeah
[00:20:19] Serena Huang: That's incredible. That's incredible. Uh, I can imagine that people are much happier as well, 'cause we need, we need tech. We need it to work yesterday.
[00:20:30] Jessica Audette: Yes, exactly.
[00:20:33] Serena Huang: And I think agentic AI, especially in a regulated industry like yours, can cause a lot of concerns for people, because when the AI is making decisions, um, without humans, what happens?
I think the job loss fears aside, there's also fear for agentic AI making the wrong call, causing harm, for example, and I'm curious if you know what made Amelia safe enough to try, safe enough to deploy even in a regulated environment like yours.
[00:21:10] Jessica Audette: This would be where I recommend you bring a sweet treat to your CISO and work collaboratively with that team.
But I think for us, what were we trying to solve for? We have some brilliant technologists at this organization that were really just being burdened by redundant work that could be automated to free them up to focus on higher level solutioning for the organization, and become more innovative instead of reactive to incoming tickets in the thousands every month or week or what have you.
Mm-hmm. We are a highly reg- regulated industry dealing with patient health information. Uh, we worked collaboratively with our security and InfoSec departments to ensure whatever we were planning to stand up would satisfy security and compliance. So, uh, that would be my first re- uh, recommendation to who- anyone who's looking at, at, at that.
[00:22:05] Serena Huang: I, I think for anyone working in AI, those are your friends, even, and, uh, a little cupcake, um, will, will not hurt, for sure.
[00:22:16] Jessica Audette: It'll go a long way. It
[00:22:18] Serena Huang: goes a long way. Is there a place... If, if someone is looking to start in agentic AI, how do you help them think about where to start? Um, is there any recommendations that you might have for them, regardless of it's, if it's regulated industry or not?
[00:22:34] Jessica Audette: Yeah, I think a lot of times people jump on the train without knowing what direction they wanna take the train in.
[00:22:40] Serena Huang: Mm.
[00:22:40] Jessica Audette: As much as we wanna get ahead of the curve in technology, we wanna be first to the finish line- When deploying AI, it's important that you have a strategy, uh, surrounding your plan. So what are you trying to solve for?
What roadblocks are you rele- are looking to alleviate? What will AI make better- Yeah ... or help you with? And so, um, anytime we look to implement something in that nature, those are the questions I recommend folks take a minute and think about. Because just deploying AI to deploy AI is, it's like having the cool kid on the block with the cool Ferrari.
It's not gonna go anywhere if you don't have the road to drive it and you don't have the destination in mind, essentially.
[00:23:23] Serena Huang: Yeah. I love these analogies. It's so true, and I, I think years ago when data is the new oil or w- you know, what- whatever that analogy was, it's almost like data/now AI have become the shiny object, and we forget the most important question is what are we trying to do?
What are we trying to achieve? What KPI will this improve? What are the outcomes we're looking for? And actually get alignment on that problem before solving it, either, you know, with more data or with more AI or m- more sophisticated AI like agentic AI. And, um, and it may not be the solution if we, if we don't even know where we're going.
[00:24:14] Jessica Audette: I would say for those of my Geeks that are into Lord of the Rings, do not become a Gollum. My pretty is not going to solve all of your ans- your issues and questions. So exactly as you said, review the KPIs. What is it going to solve? What is it gonna achieve? How is it going to help the organization? What kind of ROI are we looking at achieving or seeing back for us?
[00:24:39] Serena Huang: We started this conversation earlier about speed because that's incredibly important in cancer research, in research in general. But I'm curious if there is such thing as too fast in, in discovery, um, because we all have to be balancing thinking about how do we implement AI responsibly as well. Are there any guardrails that you can share with us and, and advise our listeners on?
[00:25:09] Jessica Audette: Well, I think I would say, and, uh, this episode is not sponsored by Ferrari, but I would say when when you're saving lives, the answer is no, right? Speed, the, the speed of the vehicle matters, so the performance of the platform matters because those turnaround times mean that we're getting treatment plans based on in- individuals, vaccinations, immunotherapies, clinical trials.
All of those things are getting back into the pa- the hands of our clinicians that much faster, which means it's getting to the patient bedside that much faster. Of course, we want to implement platforms that are secure, ethical, and really well thought out. Um, speed is really of the utmost importance when it comes to cancer care Uh, I think our organization has done a phenomenal job working collaboratively with NetSec and InfoSec to ensure that we've put the right kinds of guardrails around, uh, the platforms.
And if there's any patient health information being leveraged on those platforms, that we de-identify it, that it's further encrypted, that there are more secure protocols surrounding those particular use cases. Uh, luckily for Iris, all the information there is de-identified, so it was an easy MVP on that front.
[00:26:27] Serena Huang: Yes. I think that, that makes it so much easier. So, um, for anyone working in-- with patient data or personal data in general that is more sensitive, finding a way to de-identify helps de-risking and getting started, especially with AI.
[00:26:46] Jessica Audette: Indeed.
[00:26:47] Serena Huang: So now I, I'll have to ask this. If you could fix one thing in medical AI infrastructure, if I give you a magic wand today, what would you do with that magic wand?
[00:27:01] Jessica Audette: Hmm. Well, since you said medical, I guess production bottlenecks aren't the things I, I could fix my, with my wand. Um, so if someone could solve that, I would be eternally grateful. But I think for me and for our organization, capacity planning is the bane of my existence I like to say we have known knowns, and then we have known unknowns.
So the known knowns are our researchers, our existing labs, their workloads. We collaborate with them to understand their future needs, and we can plan for that. That makes sense, right? Then we have our known unknowns. We know we wanna hire and retain the best and brightest medical personnel and researchers, but we don't know how many we're going to hire or how many people they'll be bringing with them, if it's a lab of eight to 10 or one.
Uh, we don't know what their needs will look like or what type of medical instrumentation they'll be leveraging. Just last year, for example, we onboarded 48 new medical instruments in the organization.
[00:28:07] Serena Huang: Wow.
[00:28:07] Jessica Audette: And that just screams storage needs to me. So how, you know, how do we go about addressing that and showing the business the ROI of continued investment in the infrastructure?
I, I know I've shared this metric before, but I was able to delineate by going back into grants that were awarded to MSK and identify those that were awarded because of the high-performance computing platform available to the researchers, and I shared that over $184 million in grant funding was brought to the institution because of that platform.
[00:28:42] Serena Huang: Wow.
[00:28:43] Jessica Audette: So the ability to do their research and on that cluster, and having those metrics to translate the technology needs into the business justifications, has really helped us be able to show-- shed the light on why it's an important investment, and a important continued investment going forward for our research community and our clinicians and our patients as well.
[00:29:08] Serena Huang: Okay. 184 million? Yes. That, that is not chump change. Oh, amazing.
[00:29:15] Jessica Audette: Certainly not.
[00:29:16] Serena Huang: Congratulations. Wow.
[00:29:18] Jessica Audette: Thank you.
[00:29:19] Serena Huang: I feel so inspired by the possibilities with AI and where infrastructure plays a role. I've learned so much, and I have one last question for you. Innovation. The word innovation gets tossed around so much, especially in AI, especially in tech.
I'm wondering if you can share with us, what do you think people get wrong about innovation as we close out?
[00:29:47] Jessica Audette: This is a really great question, and I think I'm gonna be a little philosophical here, if I may.
[00:29:52] Serena Huang: Go for it.
[00:29:52] Jessica Audette: Innovation, to me, equates to change and progress, right? There was a time when we were all fearful of computers, and now they're part of our...
They're just ingrained in our everyday life, right? There's a lot of, as I mentioned, fear-mongering around AI taking over people's jobs, but I see it as augmentation to the human resources. I don't think ev- anybody's going to lose their job. It's just going to enable them to think at a higher level and work at a faster pace, if at all.
AI will be transformative in the future of work, and it'll take a mindset shift for us all to become comfortable with that. If you're familiar with the Japanese art known as kintsugi, I would say that technology, with technology, we need to get a bigger hammer to create something progressive and life-altering, and then build those pieces back up.
You know, hammer it down, and build it back up into a beautiful piece of pottery.
[00:30:56] Serena Huang: I, I love it. So for those who are less familiar, what you're describing is, um, the, the broken ceramic bowls that we see that are pieced back together, um, with, with beautiful glue, right? That's what you're re- referring to?
[00:31:11] Jessica Audette: Gold.
[00:31:12] Serena Huang: Gold.
[00:31:12] Jessica Audette: Yes. They put it back together with beautiful gold. So the whole methodology behind that is you may have started with something beautiful, but then it created something even more beautiful, and I think we have a foundation for transformative change with AI coming to the forefront, and it's been around for a long, long time.
But now it's getting this, its spotlight. It's going to take a mindset shift for folks to become more comfortable with it, and I think it's one of those known unknowns, and that makes people nervous.
[00:31:45] Serena Huang: Thank you. That is so insightful. Thank you, Jess, for joining us today. I've learned a ton. I know our audience will appreciate all your insights as well.
[00:31:55] Jessica Audette: Thank you so much. I- it was such a pleasure to be here, and I am grateful for the opportunity to discuss AI and hopefully allay some fears surrounding that.
[00:32:04] Serena Huang: Absolutely. Well, I feel inspired. I know our audience will be, too. Thanks for listening to Deep Geeks. A huge thank you to my guest today, Jess Audette.
If today's episode made you think differently about how AI gets built or powered, share it with someone who needs to hear it. Find Deep Geeks on Spotify, YouTube, or wherever you get your podcasts. Until next time.