>>Live from Stanford university on your host. Lisa Martin. My next guest joins me with many I, our corporate vice president at Microsoft, Rick Minnie. It's great to have you on the program. Thank you for having me. Tell me a little bit about your background. So you run Microsoft advertising, engineering organizations. You also manage a multi-billion dollar marketplace globally. Yes. Big responsibilities. >>A little bit >>About you and your role at Microsoft. >>So basically online advertising, you know, funds a lot of the consumer services like search, you know, feeds. And so I run all of the online advertising pieces. And so my team is a combination of machine learning in theory, software engineers, online services. So you think of you think of what needs to happen for running an online advertising ecosystem? That's billions of dollars. I have all these people on my team when I get to work with these fantastic people. So that's my >>Roles. We have a really diverse team. >>Yes. My background itself is in AI. So my PhD was in language modeling and natural language processing. That's how I got into the space. And then I did, you know, machine learning. Then I did some auctions and then I'd, you know, I basically have touched almost all pieces of the puzzle. So from, I appreciate what's required to run a business the size. And so from that perspective, you know, yeah, it is a lot of diverse people, but at the same time, I feel like I know what they do >>Right then interdisciplinary collaboration must be incredibly important and >>Powerful. It is. I mean, for machine learning engineer or machine learning scientists to be successful, when you're running a production system, they have to really appreciate what constraints are there, you know, required online. So you have to look at how much CPU you use, how much memory you need, how fast can your model inference run with your model. And so they have to work very closely with the soft, soft engineering field. But at the same time, the software engineering guys need to know that their job is not to constrain the machine learning scientists. So, you know, as the models get larger, they have to get more creative. Right. And if that balance is right, then you get a really ambitious product. If that balance is not right, then you end up with a very small micro micro system. And so my job is to really make sure that the team is really ambitious in their thinking, not always liking, pushing the borders of what can be done. >>I like that pushing the borders of what can be done. You know, we, we often, when we talk about roles in, in stammered technology, we've talked about the hard skills, but the soft skills you've mentioned creativity. I always think creativity and curiosity are two soft skills that are really important in data science and AI. Talk to me about what your thoughts are. There >>Definitely creativity, because a lot of the problems that you, you know, when you're in school, the problems you face are very theoretical problems. And when you go into the industry and you realize that you need to solve a problem using the theory you learned, then you have to either start making different kinds of assumptions or realize that some assumptions just can be made because life is messy and online. You know, users are messy. They don't all interact with your system the same way. So you get creative in what can be solved. And then what needs to be controlled and folks who can't figure that piece out, they try to solve everything using machine learning, and they become a perfectionist, but nothing ever gets done then. So you need this balance and, and creativity plays a huge role in that space. And collaboration is you're always working with a diverse group of people. So explaining the problem space to someone who's selling your product, say someone is, you know, you build this automated bidding engine and they have to take this full mouth full and sell it to a customer. You've got to give them the terminology to use, tell, explain to them what are the benefits if somebody uses that. So I, I feel people who can empathize with the fact that this has to be explained, do a lot better when they're working in a product system, you know, bringing machine learning to a production system. >>Right. There's a lot of enablement >>There. Yes, exactly. Yeah. Yeah. >>Were you always interested in, in stem and engineering and AIS from when you were small? >>Somewhat? I mean, I've been, I got to my college degree. I was very certain by that point I wanted to be an engineer and my path to AI was kind of weird because I didn't really want to do computer science. So I ended up doing electrical engineering, but in my last year I did a project on speech recognition and I got introduced to computer programming. That was my first introduction to computer programming at the end of it, I knew I was going to work in the space. And so I came to the U S with less than three or four months of a computer engineering background. You know, I barely knew how to code. I had done some statistics, but not nearly enough to be in machine learning. And, but I landed in a good place. And I came to be in Boston university and I landed in a great lab. And I learned everything on my feet in that lab. I do feel like from that point onwards, I have always been interested and I'm never satisfied with just being interested in what's hot right now. I really want to know what can be solved later in the future. So that combination, I think, you know, really keeps me always learning, growing, and I'm never happy with just what's being done. >>Right? Yeah. We here, we've been hearing a lot about that today at weds. Just the tremendous opportunities that are here, the opportunities for data science, for good drones, for good data science and AI in healthcare and in public transportation. For example, you've been involved in with winds from the beginning. So you've gotten to see this small movement grow into this global really kind of is a >>Phenomenon. It is, >>It's a movement. Yes. You talk to me about your involvement with winds from the beginning and some of the things that you're helping them do. And now, >>So I, I first met Karen and marble initially when I was trying to get students from ICME to apply for roles in Microsoft. I really thought they had the right mix of applied and research mindset and the skill sets that were coming out of ICME rock solid in their math and theoretical foundations. So that's how I got to know them. And then they were just thinking about bids at that point in time. And so I said, you know, how can I help? And so I think I've been a keynote speaker, Pam list run a workshop. And then I got involved with the woods high school volunteer effort. And I'd say, that's the most rewarding piece of my visit involvement. And so I've been with them every year. I never Ms. Woods. I'm always here. And I think it is, you know, Grace Hopper was the technology conference for women and, and it's, it's, it's an awesome conference. I mean, it's amazing to sit next to so many women engineers, but data science was a part of it, but not a critical part of it. And so having this conference, that's completely focused on data science and making it accessible. The talks are accessible, making it more personable to, to all the invitees here. I think it creates a great community. So for me, I think it's, I hope they can run this and grow this for >>Yeah. Over 200 online events this year in 60 countries, they're aiming to reach a hundred thousand people annually. It's, it's grown dramatically in a short time period. Yes, >>Absolutely. Yeah. It hasn't been that long. It hasn't been that long and every year they add something new to the table. So for this year, I mean last year I thought the high schoolers, they started bringing in the high schoolers and this year again, I thought the high school. >>Yeah, >>Exactly. And I think the mix of getting data science from across a diversity, because a lot of the conferences are very focused. Like, you know, they, they will be the focused on healthcare and data science or pure AI or pure machine learning. This conference has a mix of a lot of different elements. And so attendees get to see how it's something is being used in healthcare and how something is being used in recommendations. And I think that diversity is really valuable. >>Oh, it's hugely valuable that the thought diversity is this is probably the conference where I discovered what thought diversity was if only a few years ago and the power and the opportunities that it can unlock for people everywhere for businesses in any industry. Yes. >>I want to kind of play off one of the things you said before, you know, data science for good, the, the incredible part of data sciences, you can do good wherever you are with data science. So take online advertising, you know, we build products for all advertisers, but we quickly figured out that are really large advertisers. They have their own data science teams and they are optimizing and, you know, creating new ads and making sure the best ads are serving at all times. They have figured out, you know, they have machine learning pipelines, so they are really doing their best already. But then there's this whole tale of small advertisers who just don't have the wherewithal or the knowledge to do any of that. Now, can you make data, use data science and your machine learning models and make it accessible for that long table? Pretty much any product you build, you will have the symptom of heavy users and then the tail users. And can you create an experience that is as valuable for those tailored users as it is for the heavy users. So data science for good exists, whatever problem you're solving, basically, >>That's nice to hear. And so you're going to be participating in some of the closing remarks today. What are some of the pearls of wisdom that you're going to enlighten the audience with today? >>Well, I mean the first thing I, to tell this audiences that they need to participate, you know, in whatever they shaped form, they need to participate in this movement of getting more women into stem and into data science. And my reasoning is, you know, I joined the lab and my professor was a woman and she was very strong scientists, very strong engineer. And that one story was enough to convince me that I belong. And if you can imagine that we create thousands of these stories, this is how you create that feeling of inclusion, where people feel like they belong. Yeah. Look, just look at those other 50 people here, those other a hundred stories here. This is how you create that movement. And so the first thing I want the audience to do is participate, come back, volunteer, you know, submit papers for keynote speeches, you know, be a part of this movement. >>So that's one. And then the second is I want them to be ambitious. So I don't want them to just read a book and apply the theory. I really want them to think about what problem are they solving and could they have solved it in the, in the scale manner that it can be solved. So I'll give a few examples and problems and I'll throw them out there as well. So for instance, experimentation, one of the big breakthroughs that happened in a lot of these large companies in data science is experimentation. You can AB experiment pretty much anything. You know, we can, Google has this famous paper where they talk about how they experimented with thousands of different blues just to get the right blue. And so experimentation has been evolving and data scientists are figuring out that if they can figure out interactions between experiments, you can actually run multiple experiments on the same user. >>So at any given time, you may be subject to four or five different experiments. Now, can we now scale that to infinity so that you can actually run as many experiments as you want questions like these, you shouldn't stop with just saying, oh, I know how AB experimentation works. The question you should be asking is how many such experiments can I run? How do I scale the system? As one of the keynote speakers initially talked about the unasked questions. And I think that's what I want to leave this audience with that don't stop at, you know, answering the questions that you're asked or solving the problems. You know, of you think about the problems you haven't solved your blind spots, you know, those blind spots and that I think I want ambitious data scientists. And so that's the message I want to give this audience. >>I can feel your energy when you say that. And you're involved with, with, with Stanford program for middle school and high school girls. If we look at the data and we see, there's still only about a quarter of stem positions are filled by females, what do you see? Do you see an inspiring group of young women in those middle school and high school girls that, that you see we're, we're on trend to start increasing that percentage. >>So I had a high schooler who just went, you know, she, she, she just, she's at UCLA now shout out to her and she, but she just went through high school. And what I realized is it's the same problem of not having enough stories around you, not having enough people around you that are all echoing the sentiment for, Hey, I love math. A lot of girls just don't talk about us. Yeah. And so I think the reason I want to start in middle school and high school is I think the momentum needs to start there. Yes. Because they get to college. And actually you heard my story. I didn't know any programming until I came here and I had already finished my four years of college and I still figured it out. Right. But a lot of women lose confidence to change fields after four years of college. >>Yes. And so if you don't catch them in early and you're catching them late, then you need to give them this boost of confidence or give them that ramp up time to learn, to figure out, like, I have a few people who are joining me from pure math nowadays. And these kids, these kids come in and within six months they're off and running. So, you know, in the interview phase, people might say, oh, they don't have any coding skills. Six months later, if you interview them, they pick up coding skills. Yeah. And so if you can get them started early on, I think, you know, they don't have this crisis of confidence of moving, changing fields. That's why I feel, and I don't think we are there yet, to be honest, I don't think yet. I think >>You still think there are plenty of girls being told. Now you can't do computer science. No, you can't do physics. No, you can't do math. >>Actually. They are denying it to themselves in many cases because they say, Hey, I go to physics class and there are two boys, two girls out of 50 boys. And I don't think girls are in, you know, you get the stereotype that maybe girls are not interested in physics. And it's not about, Hey, as a girl, I'm doing really well in physics. Maybe I should take this as my career. So I do feel we need to create more resounding stories in the area. And then I think we'll drum up that momentum. That's >>A great point. More stories, more and names to success here so that she can be what she can see exactly what many it's been great having you on the program. Thank you for joining me and sharing your background and some of the pearls of wisdom that you're gonna be dropping on the audience shortly today. We appreciate your insights. Thank you. My pleasure. Who Rick, Minnie, I are. I'm Lisa Martin. You're watching the cubes coverage weds 2022. We'll be right back after a short break.