028 Edge Tech: Silicon to Software the Size of a Quarter with Elizabeth Samar Rubio of SiMa.ai

Elizabeth Samara Rubio, Chief Business Officer at SiMa.ai, discusses the advancements in multimodal applications and edge computing. How to get 10x faster and so much smaller. The size of a quarter!

She explains the significance of integrating various AI models into everyday devices, the importance of privacy and latency in AI applications, and the challenges of deploying AI at the edge. Elizabeth also shares her career journey in the tech industry with advice for the audience.

Podcast:

🗳️ Feedback? Questions? https://yourairoadmap.com/podcastfeedback

🗳️ Nominate Guests! https://yourairoadmap.com/podcastguest 

Takeaways

1. 🤖 Multimodal applications integrate various AI models for enhanced functionality.
2. 📱 Edge computing allows AI to operate on devices like phones and cars.
3. 🔒 Privacy and latency are critical factors in deploying AI applications.
4. 🔧 The silicon to software stack is essential for developers to deploy AI models.
5. 🔑 Understanding customer readiness is key to successful AI implementation.
6. 🏭 Different verticals have unique challenges and opportunities for AI.
7. 🚀 The future of AI will involve more automation and autonomy in devices.
8. 📈 Career growth requires hands-on experience and learning from mistakes.
9. 🤝 Networking and mentorship are crucial for professional development.

Bio:
Elizabeth Samara Rubio is the Chief Business Officer at SiMa.ai, where she drives global growth through innovation, new markets, and M&A. With a background in senior leadership for startups and F100 companies in AI, cloud, and computing, Elizabeth brings extensive expertise across tech, manufacturing, and retail. She is recognized by VentureBeat for her contributions to AI ethics and has spoken at Stanford's DFJ Entrepreneurial Thought Leaders forum. Elizabeth holds an MBA from the University of Texas at Austin and a BA from the University of Illinois, Urbana-Champaign, and she actively engages in community service and continuous learning.

Connect with Elizabeth
https://www.linkedin.com/in/esamararubio

SiMa.ai:

Chapters

• 0:00 Introduction to Multimodal Applications
• 5:27 Understanding Edge Computing
• 8:01 The Silicon to Software Stack
• 10:47 Applications and Use Cases of AI at the Edge
• 13:24 Navigating the Hardware and Software Landscape
• 16:20 Vertical Markets and Customer Readiness
• 18:52 The Future of Multimodal AI
• 20:37 Career Path and Insights
• 23:13 Advice for Young Professionals
• 25:14 Closing Thoughts on Leadership and Empathy
• 43:17 Summary and Conclusions

Transcript

0:00

We're building very good ML applications and now multi -modal applications that can run at the edge. That means in your appliances, soon your car, and of course for some of our industrial customers, drones and robots. Hey folks, this is Joan popping in to share a little bit about this episode. know, most of the episodes I talk to guests who work in software, who work with datasets and things in the cloud that they leverage different servers around the world to train models and make cool products. This episode is more unique in the fact that we are talking about Edge technology that also has a full stack. So when we think about Edge technology, You know, the other day I was on a train and there's wifi on the train, but it's really spotty, right? We're going through huge fields and the connectivity of getting access to wifi just disappears for long stretches of me being in the train. And when we think about edge technology, we're thinking about the something the size of mobile device, maybe something the size literally of a quarter our guest talks about having the ability to use data and leverage insights on that device. So if you're thinking about somewhere potentially very remote in a jungle, these types of edge devices can run sophisticated AI software on those devices in extremely remote places that don't have the connectivity that we do have here, like in Seattle. So as you think about the future of robotics, thinking about agriculture, There are different use cases in how we think about leveraging really sophisticated data in different places, situ, Elizabeth talks about. So I'm really delighted for you to hear this episode and really expand your mind if you are not always thinking about hardware and edge use cases. Okay, Let's dive in. Hi, my name is Joan Palmiter Bajorek. I'm on a mission to decrease fluffy hype and talk about the people actually building in AI. Anyone can build in AI, including you. Whether you're terrified or excited, there's been no better time than today to dive in. Now is the time to be curious and future -proof your career, and ultimately, your income. This podcast isn't about white dudes patting themselves on the back. This is about you and me. and all the paths into cool projects around the world. So what's next on Your AI Roadmap? Let's figure it out together. You ready? This is Your AI Roadmap, the podcast. Hello, hello. Lovely to see you. Would you mind introducing yourself? course my name is Elizabeth Samara Rubio and I'm the Chief Business Officer at Sima Cool, and what are you all doing over there at Sima? We're building very good ML applications and now multi -modal applications that can run at the edge. That means in your appliances, soon your car, and of course for some of our industrial customers, drones and robots. Ooh, drones and robots. OK, well, I think more and more people are talking about multimodal. We had a guest already from Google kind of thinking about leveraging different data sets. Would you mind in your own words just refreshing people on what does multimodal mean in this day and age? Yes, absolutely. think the best way to make it familiar is to talk about the census. So the way we see, hear, speak, read, these are coming together in the form of models or artificial intelligence models and machine learning models that we can run on edge compute devices that let people now have these models do some of those tasks. So for example, in our cars, the ability to interface with your vehicle features through voice. Those are just examples of how multimodal is coming to bear. And some of us probably have a few cameras in our cars that help us with parking or making sure that we're staying safely within certain lanes. That's another example of, again, video, and audio coming together in one device. That's awesome. Yeah, I certainly saw this in my PhD, which was mostly focused on speech recognition. When we think about how that is implemented into augmented reality products, virtual reality products, you know, using, as you mentioned, these different senses to build these really cool, innovative products that are coming to them. yes, it will be. I'm convinced that multimodal will be far more impactful in how we interface with AI in our daily interactions than anything we've seen yet. Hmm, that's beautiful. I agree. Well, and you also use the term edge, which is not the side of a piece of paper. But when people think about because right now we're on Wi Fi, right? This connectivity that most people in cities feels omnipresent. But that may not be the case in certain contexts. Would you share what is edge for those folks who don't think about that term? Okay, so Edge definitely has, a family of devices. Edge could be AI models running on our phones, Edge could be running on your laptop. Edge can also be running in your vehicle. But it also means running on, a workstation or a server or what they call industrial PCs. Edge can also be a healthcare device. where you're embedding AI for purposes of improving the procedure, i .e. surgical procedures. It could also be embedded AI that goes into machine equipment. So in industrial sector, you may have like a welding machine where instead of looking for applying AI, for doing a of a review of work completed, AI is getting embedded into the equipment itself so that we can actually do an in -situ analysis of how the work is getting done. done so corrective measures can be taken. So these are all examples of different edge devices. And what we're doing with our silicon and our software at Sima is enabling those devices to be able to carry some of these out of AI AML models. And again, as we mentioned, we'll talk a little bit about the new MLSOC, or machine learning system on a chip, that we launched last week that now introduces multimodal and GenAI capabilities at the edge. Congratulations. Launches are such a big deal. Yeah. Well, and as you think about, I am learning more about your company, are you mostly on the hardware side? Are you also in the software? Do you think about the whole stack? Like how do you all think about where you sit with your products? Very good. I'm going to ask two questions in two parts. What do we do and what do our customers need? So what do we do? We have designed a silicon, which we call the machine learning system on a chip, and a software that enables customers, i .e. developers, to be able to port their trained models onto our MLSOC. That's what we do. I call it the silicon to software stack. so that you, the developer, can go ahead and deploy your Edge ML. That's what we do. What do customers need? Well, that's just the beginning. There's a full stack that customers are looking for when they're looking at solution systems, all the way from the presentation layer, which is what we see on our screens. Right below that, all the analytics or what we consider business logic that makes sense of all the inputs that's being provided to the application. And then One part of that in our applications today, we enable Vision AI. So in our case, we're part of the stack that provides that stack a prediction as to what did we see, what is it, where is it going, or how many there are in there, is it good or bad. We provide those types of predictions or inferences to this stack. Customers buy the stack. Yep, that makes sense. Well, and as one of my customers is an agriculture customer, so I've learned far more about IOT on the farm or like thinking about humidity or there's different, even though the backend to me looks extremely similar to other builds I've done in contact centers, the types of data, humidity, looking at the plants, triaging things, it's fascinating to hear some end customers talk about, I just want a dashboard. Like if you could just give me an overview of my system, whereas the data lakes. Yes. are some of the long -term value for the investors who see optimization. We're like, how people think about what is valuable in the moment. But also, as you mentioned, the whole stack, as well as security of as these data sets can be tremendously impactful. we need to just data sovereignty is such an important thing to think about writing into our contracts, making sure that our own companies are protected. I love that you brought that up Joan, because if we talk about not just what is Edge and what we do, but why do customers choose to deploy AIML at the Edge? You've touched already on some of them. Privacy is a big concern. So one of our potential customers that we're talking about, they do exercise equipment that people buy for their homes. The way they interface with the equipment is that they do capture an image of the person. So the number one thing for them is we need high performance AIML to be doing real time image capture of the person and what they're doing. But at the same time, we cannot have that data leave the system for privacy reasons. So that's number one, to your point. And then there's obviously the other one, which is latency. Some of the applications that we're working on with customers, we literally have 800 milliseconds. to respond. So that's under a second to go do a lot of work and ultimately take some form of action, physical action. You're not going to be able to do that when you have to send the workload to the cloud and back. So we call that ultra low latency applications, which is the second why. And then, of course, we all know a little bit of the cost of running inferences. in the cloud versus the edge, I think you're going to gain an appreciation that there's definitely a business reason for the commercial side of moving to the edge. The cost is definitely a very, it's a whole different ballpark. Wow, that's fascinating. when we think about latency, and I'm thinking about for people who aren't as familiar with the builder side, just like the speed, like when you use ChatGPT and it takes a long time to get your response, right? That latency becomes a friction. You're like, should I pop over to my email? Like, I, you know, how long am I gonna be waiting for these answers? As you mentioned, all the computing behind it, and even potentially a content reviewer, a human. taking a few seconds to review the answer, think people may forget or they're just like, this is slow. And we're like, we are working so hard to give you excellent results over here. It's Jonah, I'm smiling because just the other day I was working with a customer. We're about to do the third project with them and we are shaving. We started with shaving two seconds, one second. We're down to like talking about 15 milliseconds. It's literally, it's like, wow, I feel like this is the Olympics. Everything is measured in such a small increment of time. to deliver that user experience you're talking about because you don't have that moment of wait. And that's, think that's going to be the kind of expectation that, especially as we introduce more, I'll call it payloads for now, more payload at the edge with multimodal. People don't change their expectations in terms of how much time does it take for the system to get back to you. It should be just as fast as it was last time or better. Absolutely. Well, and I think as, right, things are getting faster. It's funny, they're like, they want it faster. They want it faster, the Olympics. I think all the optimizations in builds I've done, people are like, the launch is, like, we're done. And like all the maintenance and optimization and the cost scaling as these data sets potentially get bigger and bigger, that's maybe one of the first steps along the way. agree. there's something, you know, we've to your point, when we start projects with customers, you know, there's a lot, mean, everyone has a lift, you know, there's not a lot of, there's a lot that has to go in and making a successful production ready system. And so when we get to the point where we've done a POC check, good. Then we get, we're getting ready for a factory acceptance test or a FAT. And that takes not days, but a couple months, right? Check. And then you get installed on the production line and there's a validation period for a couple months. So at least eight weeks. Check. And everybody's tired. And you're like, no, that was just the beginning. Now you get to live with it for the rest of your planned life. it's in these types of applications that I'm describing, it's a minimum of 10 years. Yes. So that, you know, this is not a, I'll change it out in two years. It's like, no, it'll be around for a minimum of 10 years. So we're putting not just system in place, we're building the, helping the customer build an infrastructure that lets them manage something for 10 years. And part of that infrastructure is MLOps. I call it EdgeOps because MLOps, there's a lot of things that may go back to the cloud, but in the Edge, you have that, you have that extra element of you also have to take care of the actuation, the physical things that you're doing with this model for the next 10 years. There are physical things taking place because of the inferences and that means you have to include the devices as well. Yeah, I think being in Seattle, I'm just surrounded by software. So the idea that something would be around that long, I've only done a very few hardware projects. like, I did a medical hardware and they're like seven to 10 years just to get it to market because of FDA approvals and legal. like seven to 10, like the bets you have to make, right? Those investments of time, capital, et cetera, are, it's just different world, different, different. world and we almost have like two types of projects we do. We do process and product. Product is like what you just talked about where you're embedded. The Sima is embedded into the physical product. That means we have to work through the life cycle, whether it's automotive, which has a similar type of life cycle in terms of getting designed in, or medical devices. Those two are good examples to what you're saying. It takes a while. So it's a product. go to market. The process ones can go a little bit faster. We're not waiting seven to eight years, but it's going to take a year to go from where I said the POC were good. We have a feasibility check mark on several KPIs, but taking it all the way to production and saying good, pass the validation period, and then launch for 10 years, it can take up to a year. Yeah, well, and that proof of concept, right? Even believing that it's possible. I saw a startup a few years ago with pacemakers and thinking about cybersecurity, which blew my mind that you'd put that in your body and it might get hacked. was like, that's a problem I hadn't even thought of. But when you mentioned drones, are there certain verticals that you're really thinking most about? Do you look across? How are you thinking about that market? yes. So we really have eight verticals that we're thinking we've been targeting. Of course, there's a priority order because there's a level of number of customers that are ready. That does vary from vertical to vertical. So the first one is industrial. So the industrial has had a J .I. for a little bit already. And before a J .I., they had no more of the legacy. machine vision algorithms that were already doing things like process monitoring and quality inspection. So this vertical is not new to the automation that comes to bear with when we release these types of AI applications. Then there's smart retail. So this is more of not just doing the tracking of foot traffic in and out, but is also being able to do unique identification. of persons or objects in the scene and then being able to map that to some form of logic around them basically ends up being analytics. What can I do for you? As one of our partners said, who's one of on the vertical for smart retail, they're like, we just want to do what Google Analytics did for retail, but we want to do it in the store. I was like, OK. So that's what we're enabling. And a lot of that level of work, there is a very different profile in cost when you bring it to the edge. And of course, you can continue to protect the privacy of the people you need to protect who are part of your consumers. The third one is going to be aerospace and defense. If you can think of something so small, something so powerful, and something with very little power requirements, The sweets are small mobile things. So drones and for purposes of either change detection, anomaly detection is a big demand there. It's a good fit. Last but not least, agriculture is definitely something we're looking into on the product side and then the healthcare as well. So that kind of covers the breadth of what we've been looking for in the order. I would say in the order that we've been able to go to market. based on customer readiness. With multimodal, we're looking very similar at same markets, but we're looking at a new form factors. We're getting with our Modalix family that we just announced last week, our chip is a third smaller. And the improvement is over 10x. where, and the power envelope is, we got on the map, one of the things that put us on the map was 50 taps at five watts. I know those are big terms, but equate 50 taps to performance. How much can you do on this SoC at five watts? And if we look up five watts, look at how much. power is required for a light bulb. That's what we're talking about. Doing a lot with that level of power. And now we can do that even more. And we have to because multimodal applications will require more performance. Period. There's no way around it. is super efficient. and how like the size of it as you're showing those of you who are listening can't see Elizabeth with her hand showing me like what what size are we talking about for these type of yeah. launched is the size of a quarter. Yes, so you have different form functions. So one is a customer can buy the chip and they can build a board around it. Or we will also be offering what they call a board or half height, half length board that can go into an industrial PC or a server for purposes of supporting the edge. of compute as well. And we'll have many more form factors coming soon in the next, I would say, three quarters. Okay, y 'all are moving forward fast. Well, how big is the team today? How long have you all been, speaking of timelines, how long have you been working on this company people building it? so Sima we're going to into sixth year. And then a number of people, we're approaching well over 160 people, a good mix of ML specialists, silicon individuals, so people who've done this before, and then software. So you have quite a bit on the engineering front. And then on our commercial side, we want to get more people with systems level experience and domain or vertical specific experience. One of the things I learned at AWS was for AI, you have to go vertical. people, technically, the models tend to perform when they're far more. They're trying for specific domains and tasks. We know that. But also from a customer point of view, very few retail customers are interested in hearing our use cases for industrial. And likewise, very few industrial customers are interested in smart retail. good. I mean, those variables are different. It's interesting to hear that. Hmm. Yeah, one eight verticals. I mean, did you all I'm just thinking from a business standpoint, Well, yeah, I'm just thinking about, you know, six years is not that old for that much traction. And how do you connect with companies? It's interesting to hear that industrial was one of your first ones, because I've heard no offense to people in the industrial industry, but some stodgyness about are they ready to adopt things? This is how it's always been type of narratives. Did you? Did the company already have pre -existing relationships? Is there eagerness that I don't know about? Like, I'd love to hear, yeah. Yes, yes. I would say yes to all of them. The best way to say it is, and I'll talk a little bit like marketing or business, you have to know how to qualify customers. So we qualify customers based on maturity, which it tells us like, how ready are you? And we have a few questions to get that to kind of put a customer on the spectrum. And then number two, what's working or not? What have you tried and it didn't work or what are you using today that's not meeting expectations or what have you tried or want to try that you don't think you can do with what you have today? So that's number two, a sense of like, where's that white space that we can possibly talk about? And of course, like any opportunity, the sense of urgency. There's a lot of interest in POCs. And we ask questions. I mean, there's nothing wrong with the POC, but every business should know if they're stepping into an opportunity that is POC or it's POC to production. And you can ask a few questions such as, can you walk me through your operation schedule? Can you walk me through your operating budget? Can you please introduce me to someone who's going to be the one whose P &L is going to support the next x number of years of this project. And then you get a sense of where are you in the process. But no, but to answer your question directly, we follow those three steps to know who an industrial is ready to kind of move on to use an ICE example, eight cylinder, not four cylinder type of cadence. And we find those, they're there. The thing is having In industrial, you don't want to go in and say, let me tell you about my chip or let me tell you about my software. Nobody wants to talk about that. They want to hear this. Tell me about what you've done and tell me from a systems level. How long did it take? How much did it cost? What were the good parts that worked out, which were the hard parts that you had to kind of work through with that experience? If the person has the experience and we do. We have people, myself included, who put systems into production in the manufacturing sector. That brings a level of confidence that, okay, you're not just going to come here and push chips. You're helping me think through the system and you understand where I am in pushing the limits of what we do already. So there's definitely, I call it pockets of possibilities and every vertical. Absolutely. No, that's so cool. And I think as I, I'm from the technical build side and I'm certainly, you know, sales is a growth area for me. and really thinking about that, that listening and not the pushing up. is the answer. really building that relationship, building that track record of here's what we've done before. Here's how that may or may not map on. And I think, I've certainly like leading with AI or just throw AI on it. yeah, exactly. has not been working out. But I've also, I the mindset of stakeholders, I've certainly taken meetings where people are like, someone forced me to take this meeting, I'm just checking it off, and I met with some AI person, or just, I think the maturity of different markets and stakeholders who might be older and kind of ready to phase out their career rather than taking on a new five, 10 year project is just not of interest necessarily. it seems like so much has been done, but also, you know, it's amazing what you're gonna see next. Not as about your company, just your When you kind of project out of your part of the field, what do you think you're seeing? Like really thinking in a broad way about edge, about multimodal. you're on the ground doing the work. So I'd love to hear what you think is on the horizon. I think it's with the multimodal capability, what people have been talking about with AGI is going to become more expressed, if I would say that. You talked about your early example with a agricultural company. Imagine a device that now can make an inference or prediction based on how hot, what's the weather, temperature, humidity, condensation, precipitation. who is around me. What did they do? Where am I? All those different questions are now and will have models behind them answering them in real time. And there will be an inference made. So what I see, what's going to happen is the AGI will probably accelerate because of multimodal capabilities. All of a sudden, you're going to have a data set that doesn't really exist yet. And people will push inference to the edge because people will want more automation or more autonomy of physical devices or of tasks. And that's really where it's going. I think when given that trajectory, Joan, safety, trust, privacy. And ultimately, a respect for the individual that's going to be, that all of us are going to be impacted. think that elevates even more the importance of us taking the time to think about the policy implications of what we're enabling. Absolutely. I liken it to an electric car or self -driving car where it's leveraging all this data. It's been going through all this testing to even be out there. But people were like, do I want to give it the wheel? Is there a human in the loop that can stop the things? really thinking about, as you mentioned, with these predictive things, these rich contextual understandings, that these end customers are like, we could know that. Like even timing the market about when to sell certain produce, that can be a 20 % margin difference. Like this is, it's wild to me that that's just one piece, as you mentioned, of these whole AGI, so sophisticated and beyond human capabilities. Yes, we're going to create, variable, we're going to help create the datasets for that. Right. know how powerful data is and we've talked on this podcast, but like data being one of the key foundational layers for any of the AI things we want to build. And I still think, you know, basics of data cleaning and data engineering are still undervalued on the market. Or I don't know. Yeah. They absolutely are. when you do a Gantt chart for a project, the thing on the bottleneck, right there, it's called data. Almost everything else can go as fast as you want it to do, but that part right there, data, that, take your time. mean, it's the one thing you just don't rush. love that. I love that. Okay, well, as we're thinking about the field, I bet there are people listening to me like, whoa, what a cool job Elizabeth has or what a cool field. You I might be interested in pursuing that. Would you mind sharing as far back as you wanna go, how did you get here? Well, first, just banner point. I just always looked at the horizon of what's partly just who I am, but also in the earlier in my career, it was part of what I had to do. I had to create frameworks. I had to do research to try to put a North Star, or at least a hypothesis of what is a North Star. And that just built a muscle that I just continue to till today, right? Knowing how to ask the questions, knowing how to do the research, knowing how to make sure you're not doing too much confirmation bias along the way. But then given that that's natively what I do, it helped me look at opportunities through a lens of what do I get to do so that I can make that happen? So it always boiled down to impact. So for example, let's go back to the first time I actually touched natural language processing was 2001. People have no idea that NLP was around in 2001. I'm like, yeah. Neural networks were, MIT was working on it. Stanford computational linguistics department was working on an alternative to neural networks. It was more language -based type of knowledge basis. for, So it's been around, it was contact centers. I'm like, 2000, 2001, it's been around. So I started there. It was a company called YY Software and it was out in Mountain View and loved my interview with the head of engineering and one of his teammates. It took two hours, we were there, it's like 8 p Like literally just walking through like the construct of a knowledge base and how the architecture set up to be able to do the language analysis. It was crazy. And ever since then it was like, okay, well, where is this going? So then I got into the, opportunity to work a few years later into machine vision, which is more of a legacy, but it was all of the mathematics, all the algorithms. to start putting automation on really fast lines. mean, some of the lines we would put vision algorithms on were moving at 2 ,000 meters per minute. Put that into miles per hour, that's over 70 miles an hour. And we were looking for things like the size of a grain of black pepper. So it's like, you just catch it. It was, as you could tell, mean, that was amazing. Like you go into these places and yeah, they're. You probably don't want to wear your best suit there, but they're hardworking people doing hard things. these machines are just blow your mind away what they can do. And now you're to put a vision system on them. So that was the next level of leveling up on the technology. Did a lot of work in other areas like central. did some work there with digital manufacturing, digital twins and threads, embedded AI into smart products, consumer products, and then went to AWS. Throughout this whole period, I got bored a little bit. And I said, I want to go learn more. So I went to Stanford. And I took, I think, like four courses in Python and just started learning how to code some of these. models and I wouldn't say I will not be a PhD level, but I do know enough to understand what's going on in the applications that our builders are putting together here. So that is a little bit of who I am, but also who I am natively allowed me to take some bets on my own career. No, and these bets were towards how do I impact something that's coming? That's amazing. Well, and I love hearing this curiosity and the, bored, let's learn more. Like, let's go back to the, that's the reality, right? I mean, we go through our careers. Well, and how did you, I mean, natural language processing back in 2001 is pretty darn cool. How did you get in the door? Did you get a math or linguistics degree? Like, how did you make those first steps? Yeah. before I went there, I was at Hewlett Packard and I was a global product manager there. So I had already, it was a horizontal offering, but we were beginning to create programs on a horizontal capability and doing some vertical. So the work got a lot of visibility. Long story short, I was contacted through a recruiter and they're like, Basically, we have this great horizontal thing and we just don't know really where should we point it to. So like, which markets should we go after? How do we go after those markets? Who should be our partners? That's how I got the foot. That was the foot in the door. It was a recruiter looking for, look, we have these great scientists, we have these great developers. We just don't know where to point the company to. And that's where I was brought in. And the rest of was curiosity to use your reference. You just have to spend the time with the engineers. I there were many nights where I just stay afterwards for two, three hours and we just, they just whiteboard how they're thinking about their architecture. And it was just me learning. Wow. Well, and even there's so many, I mean, you lived it. but like, I'm still trying to go back to like, I guess those were really early days of Hula Packer. HP, yeah, yeah. But thinking about their different product offerings, being, when I'm thinking about product, how does NLP, but like a recruiter reaching out to you that you're a high performer and that there's just opportunities. to upskill, is that really what I'm hearing? Okay, wow. Joan. I had like one. Almost three promotions in three years at HP. Yeah. Did you go to college? Did you get undergraduate degrees? University of Illinois Urbana -Champaign. And then I got my MBA at UT Austin. And I was very fortunate at UT Austin at the time that I went there. They were piloting this program where you can do a dual track, one in MBA school, but also IT. So it was thank goodness that you can do the technical part with the business part. And you can see these two roles coming together because I believe if I go back, 35, 40 % of the MBA class that I went to was technical and wanted to go to commercial. And for the rest of us who are going to commercial and wanted to get into the technical field, we needed that, let's say, bench building, if you will, on the technical front in order to be able to step into that role. totally. Well, I think that's what an innovative program. My cousins, I have some that are like 10, 15 years younger than me and they're going to college and there's emerging technologies is a degree under the business school. back in the day, that was not a college major. I've been in the R &D side where we can build stuff, but can it get sold? You know, two stakeholders who are making all these fancy charts and looking on these models and they're like, where's the bar chart? What is the ROI? How does this match the dollar signs? And I'm embarrassed to say that really opened my mind to, gosh, technical pre -sales. How does the customer really understand your work? Or is sales pitching something that we can't even build yet? We're like, there has to be a fortuitous conversation that's happening. Yes. The technical pre -sales, you need the person, one, the technical foundation. That's the table stakes. But you need someone to, like you said earlier, that empathy. The customer you're taking with is trying really hard to understand not everything you know, but what you're trying to convey, right? And the questions may not come out correctly or may not be complete. So having the patience, knowing that the customer is spending time with us because they believe so far what they've seen and they really want to accomplish something. So that pre -sales technical person has to have that sense of empathy that this is a, I mean know it's a little bit of overused words, but there is a journey. There is a change process that happens when you work with someone who's, especially with a startup. there is going to be a metamorphosis on both sides. And so you have to have that empathy. that's so beautiful. something I learned from a mentor relationships, relationships, relationships. That's how business actually gets done. I was like, really? Two years later, I was like, brilliant, But thinking about, I mean, it's really a relationship, a partnership, that empathy, listening, the trust building. And at the end of the day, there are probably going to be at least two signatures on that contract. right. And so I think that humanness, even when we're doing these super sophisticated builds, cutting edge technologies and performance, but I think you're right. That savvy on both sides. There's some salespeople like, I did not appreciate you enough. I'm in the past of how hard it is on that side. Well, when you think about folks today, I don't know if you get them in your DMs too about like, your career looks so cool. They might ask for advice, recommendations. Is it about a degree? Is it about different skill sets? Is it about mentorship? I'm just throwing things out here. But when people ask you, know, young grads, people pivoting, what advice might you give these folks? I would say young grads, get in, get your hands dirty, show. Really fast. Like just show. And the best thing of being right out of school is you can, you know, choose your manager carefully. That's always number one. Where your progress, your results are going to be measured by how many mistakes you made. And let me tell you why I say that. If we're not making mistakes, then we're not trying, we're not going fast enough or we're not trying new things fast enough. Pick your managers wisely. Pick someone who through their actions, their career, they've demonstrated what they're asking you to do. Right? So move fast and make mistakes, but as long as you're making mistakes in that direction where it's like moving forward, that's called a lesson and we all needed that. And so let's keep going. So look for that in that person. For someone who's trying to do a pivot, it's worth network, network, we're network with the people that you want to be working with. But at the same time, if it's like a 90 degree pivot. Take the classes. As I mentioned, Stanford was amazing. I really love the continuing education courses. You know, it's a little hard when you work all day and then you go at night and then you have homework and assignments and you have a test. But it's worth it. I can't say enough. It's worth it. Go get it. Go do it. It's worth it. And I would stay away from like, certifications that say you're AI certified, I would say, you know, go to a class where you actually have to like code. So that gives you a little bit more of a, because again, whatever pivot you're doing, people are going, you're probably going in a technical field and you're probably going to, in a technical field, people really appreciate show. Absolutely. I think that's pretty much other types is like sometimes same field, not a strong pivot, just maybe the next step in their career. Of course, common denominator network, but offer to help that person. Yeah, I just had one last yesterday and she's like, well, is there anything I can help you with? And I was like, given her background on, she's a COO at another company. I give it a better, I'm like, you know what? I have some infrastructure that I've been investing in and we need the operational execution to make that infrastructure pay off process people. I have some ideas and said, well, if you like, I'll give you my thoughts. That was just a simple conversation and we're going to talk again in two weeks. Hmm, absolutely. it has to be a two -way street. But we really offer each other our different zones of genius. And I think that's that opportunity of how can I help? Can I make an introduction? Like as we participate in these ecosystems and network, and I completely agree with you. having the hands -on experience to say what you've actually done, get your hands messy, but be in those rooms to meet and bump into those people who eventually may want to be hiring or switching, try to be in those rooms. Yes, absolutely. I think you bring up a good point and this was something I learned not too long ago. The difference between mentorship and advocacy. So mentorship, okay, absolutely it has to be bidirectional. But advocacy is someone who in a room or in a discussion without you having to be there will stand up for you and say, have you thought about this person? Or have you thought about this rule differently because you need to and I know I have the right person that you need to have some time with. That's advocacy. And those are even more rare, but extremely valuable. Absolutely, know, people saying good things about you in rooms you're not in. that's some of the best referrals I've gotten. Someone said, your customer raves about your work. Or like, yeah, I was like, damn, I didn't even ask for it, right? It was just something, it was a conversation, a room I wasn't in, but vouching for me to someone I didn't even know whatsoever. And that, gosh, just, hits me hard in a really good place that that work pays off, that relationship, you know, is genuine, has been built over time. Yes. So I think that spans one's whole career potentially. Are there any things that we haven't mentioned that you would want the listener to know takeaways from this conversation otherwise? I would think that one of things I talk about with is clock speed. Know your clock speed. And then know how to read somebody else's clock speed. And then three, if you're in a position of setting direction for your company, know how to read the clock speed of the market. And I think that that's a good way to close is it really comes down to how fast do you... Where do you feel comfortable at? At what pace do you feel comfortable with? And Joan, I'm picking up from both of us. We probably are very comfortable at 100 miles an hour, right? Yeah, we're like, whoop. But being a leader means winning with a team. And not everyone should be running at your clock speed. We need a diversity of clock speed. that means be good at knowing how to do that. And I think that would be the best thing to say because ultimately, whatever things we try to do in our lives and our careers requires change management. and change management is going to come down to understanding the clock speed of others. that's so brilliant. Well, that's kind of the relationship and empathy or that learning, knowing you're the people around you to build. That's, wow, that's I've not heard that one. And it's so beautiful. It reminds me of I don't know if you've read the Buy Back Your Time book, but it's thinking very effectively about going from a scrappy startup to a more polished org and all the relationship team stuff that goes on about time and effectiveness. And there's one person called the speed demon that is driving their team unhappy because of all the fast decisions and not necessarily getting buy -in or communication. I read it and I was like, no, my team has definitely said these things to me and how I can be a more effective communicator, get buy -in, know, hear concerns from my team is, like, or just as I hope people are hearing as a leader, continuing to learn and grow to be the most effective leader that I can be for any team. So hopefully other people can listen to that and be like, you know, fast speed, also I think that, learning and growing along the way as humans. We have to. element. know, people, when somebody wants to ask me in an interview, so do you just go fast, break the rules, and break things? And I'm like, no, if you do that too often, you might end up alone. So you can go fast, of course. That's your clock speed, whatever that fast is for you, right? But to your point, but be human along the way. Absolutely. Well, that's a beautiful place to end. This has been a lovely conversation. And if listeners want to find you, follow your work, Sima, et cetera, where do they need to go? Please find me on LinkedIn and I'll be happy to accept your invitation. Thank you so much. We'll have a wonderful rest of your day. Thank you. Oh gosh, was that fun. Did you enjoy that episode as much as I did? Well, now be sure to check out our show notes for this episode that has tons of links and resources and our guest bio, etc. Go check it out. If you're ready to dive in to personalize your AI journey, download the free Your AI Roadmap workbook at yourairoadmap .com / workbook. Well, maybe you work at a company and you're like, hey, we want to grow in data and AI and I'd love to work with you. Please schedule an intro and sync with me at Clarity AI at hireclarity .ai. We'd love to talk to you about it. My team builds custom AI solutions, digital twins, optimizations, data, fun stuff for small and medium sized businesses. Our price points start at five, six, seven, eight figures, depends on your needs, depending on your time scales, et cetera. If you liked the podcast, please support us. Can you please rate, review, subscribe, send it to your friend, DM your boss, follow wherever you get your podcasts. I certainly learned something new and I hope you did too. Next episode drops soon. Can't wait to hear another amazing expert building in AI. Talk to you soon!