image: thumbnail.jpg
Forecasting AI Demand at Microsoft | Sajay Suresh | Data Science Hangout
We were recently joined by Sajay Suresh, Senior Director of Data and Applied Science at Microsoft, to chat about data center supply chain planning, forecasting AI demand, and navigating data science careers.
In this Hangout, we explored how the emergence of technologies like LLMs changed projections for data center demand. Sajay discussed how forecasting for something with little historical data, like AI demand, required drawing analogies from the past, such as comparing the training/inferencing model to the iPhone and its App Store. A major complexity in current supply chain planning is the lack of fungibility with modern GPUs requiring specific infrastructure like liquid cooling, meaning data centers designed for GPUs cannot easily be repurposed for traditional compute/storage workloads, increasing investment risk if demand is lower than planned.
Resources mentioned in the video and zoom chat:
🔗 LLM Workflow Demo with Joe Cheng → https://pages.posit.co/05-28WorkflowDemo.html
🔗 Posit::conf 2025 Virtual Registration → https://posit.co/blog/posit-conf-2025-virtual-experience-registration/
🔗 Sajay Suresh on LinkedIn → https://www.linkedin.com/in/sajay-suresh-12687631/
🔗 Find mentors on ADPList → https://adplist.org/
🔗 Officeverse R packages for Office documents → https://ardata-fr.github.io/officeverse/
🔗 Microsoft team meetup video on capacity planning → https://www.youtube.com/live/07j22d4B_hA?feature=shared
🔗 Seattle Data And AI Security community → https://www.linkedin.com/posts/seattle-data-and-ai-security_microsoft-fabric-tour-seattle-data-ai-security-6891902675280633856-xLw3?utm_source=share&utm_medium=member_desktop
🔗 Quarto Gallery → https://quarto.org/docs/gallery/
🔗 Quarto Guide → https://quarto.org/docs/guide/
If you didn’t join live, one great discussion you missed from the zoom chat was about communities and meetups recommended for networking and learning in data science. Participants shared various groups like R-Ladies, Data Book Club, local tech meetups, and specific conference recommendations like Shiny Conf and DataConf.ai NYC. What's your favorite data community?
â–º Subscribe to Our Channel Here: https://bit.ly/2TzgcOu
Follow Us Here:
Website: https://www.posit.co
Hangout: https://pos.it/dsh
LinkedIn: https://www.linkedin.com/company/posit-software
Bluesky: https://bsky.app/profile/posit.co
Thanks for hanging out with us! 💛 Subscribe to posit::conf updates: https://posit.co/about/subscription-management/
Transcript#
This transcript was generated automatically and may contain errors.
Hey there, welcome to the Paws at Data Science Hangout. I'm Libby Herron, and this is a recording of our weekly community call that happens every Thursday at 12 p.m. U.S. Eastern Time. If you are not joining us live, you miss out on the amazing chat that's going on. So find the link in the description where you can add our call to your calendar and come hang out with the most supportive, friendly, and funny data community you'll ever experience. Can't wait to see you there.
I would love to introduce our featured leader today, Sajay Suresh, Senior Director of Data and Applied Science at Microsoft. Sajay, how are you today? Can you tell us a little bit about you, what you do, and what you like to do for fun?
Absolutely. Thank you, Libby, and thank you, Libby, Rachel, and Pawsit for having me on this forum. Can I say one thing before I get into it? I love the community vibe over here. It's different than most talks or conversations I've attended. I feel like people are a lot more comfortable over here, so congratulations on building an amazing community.
Oh, thank you. It's everybody. Everybody is doing the community building at every moment. Oh, hi, Geraldine. It's so good to see you. Hey, okay.
That's awesome, right? Yeah, so a little bit about me. I'm at Microsoft. I run an Applied Science function at Microsoft. My team essentially runs cloud infrastructure planning. So we are a bunch of data scientists and software engineers who create forecasting models for cloud and AI and figure out how to get the data center infrastructure for that in place. So I'm hoping it's touched your lives in some way or the other. If you have ever used a Microsoft product, be it Xbox or Windows or Azure, AI, whatever you have used, hopefully my team has had some role to play in it. So I'm excited to be here.
Awesome. What do you do outside of work, outside of data and AI stuff for fun?
You know, there is a world I had pre-kids and post-kids. So pre-kids, I loved soccer. I'm an avid Manchester United fan. So I like to stay unhappy in life is the way I call it because I'm a Manchester United fan. But post-kids, anytime I get, I just find, you know, I look for time to stare at a wall and blank out. That's my favorite pastime. I kid you not, and my wife hates it, but I enjoy that.
We all need to stare at a wall every once in a while and disconnect from reality. Well, I'm so excited to have you here and to get the chance to ask you a bunch of questions. I was hoping that you could tell us a little bit, just a little snippet of your background and how you sort of got to data science and why data science for you.
Career journey to data science
Oh, that's a great point. You know, my career started around 15 years ago, right? It was a company called Mu Sigma, which was trying something called data science at the time. Data science wasn't really obvious as an industry. It would be surprising for a lot of folks over here, for y'all younger in your career, but data science, it wasn't really clear, but there's going to be a niche small industry or a really big industry that every company needs. So it started back in Mu Sigma 15 years ago. I spent two to three years with them where I myself got introduced to the idea of data science and how decisions are taken using data.
Post which I moved to BCG and BCG was interesting because BCG is a Boston consulting group. As most of you may know, it's a big management consulting firm, but data science was interesting for them, right? It's not the most obvious place for a data scientist to be 10 years ago, but that is exactly the reason I joined them because they were trying to build out their data science practice. What they had was data science as a small service with them, but the idea was to bring folks who are well aware with data science and consulting specifically to build out their data science practice. So it was amazing experience at BCG to not just, you know, be with fortune 500 companies trying to solve strategy problems, but also have like a startup within a large company and try and build a startup out. So now if a lot of you folks are aware, it's called BCG X. I think it's rebranded itself from BCG MR to BCG X and it is a separate entity within BCG separate from the business management consulting. So that's where I spend a lot of time. First that, that was around two, two and a half years, I believe. And that's when I consulted for Microsoft and saw that there was an interesting role in tech, which was around the data center supply chain. So seven years ago, I switched over to Microsoft.
As a data scientist, we were a two member team trying to figure out how to even plan for data center supply chains. Just to give you folks a sense of what a data center supply chain is, right? Imagine if say I'm in Seattle, right? And imagine if I want a new data center in Seattle and that's what I need if I need to deliver cloud and AI, right? Cloud and AI basically means there are servers housed in data centers. It would take me three to four years to get a data center life. So as much as cloud and AI is instantaneous data centers are not because you got to get the land for it. You got to get huge amounts of energy for it, ensure it has all the resiliency needs your customer needs and it is construction, construction of buildings, ensuring there's energy and network over there. It takes three to four years. So my team essentially is trying to predict what the world would look like three to four years down the line so that we can make the investments today and Microsoft can capitalize on opportunities for their customers at that point of time.
So when you see AI being big today, I bet I have caught it three years ago. Otherwise, we don't stand a chance of serving it. And as most things in forecasting, I kind of caught it and kind of didn't catch it. That's what happens in forecasting. So that's a little bit of my journey and how I got here, Libby.
And as most things in forecasting, I kind of caught it and kind of didn't catch it. That's what happens in forecasting.
Just how you got here, what your background is, maybe what you studied, but I think that it's good to call out the things that would be good for questions. So as everybody is sort of writing their questions into Slido or thinking about them, what we want to talk about today, some of the things that we could talk to Sajay about are, like you said, how supply chains work with data science. I come from a supply chain study background, so that's exciting for me. But also sort of the navigation of data science careers. It's really interesting starting in consulting, and it sounds like you consulted for Microsoft and then went to work for them, which is an interesting transition I would love to hear about. But also forecasting and optimization. You could talk about hackathons. You could also talk about early career topics from a manager's perspective. So lots of stuff to talk about.
If anybody is interested in all of those things, I encourage you to hop in. I think that you also mentioned to me quick iterations, quickly iterating an MVP solution design. So I'm curious a little bit about that. But we do have some questions coming up in Slido, so let me go take a look at those.
One other thing, Libby, as you look at those, one other perspective is I started in consulting data science, but spent a lot of time there. And then I think at Microsoft, I grew into product data science. Data as a product. So if you want perspective on what you are doing and how you want to change or do not change what business models are suited for consulting versus product, happy to talk to you.
Evolving as a data scientist at Microsoft
I think that's a great question. Well, I have a question from Noor that I really like. Noor, if you are available, we will find you and unmute you so you can ask that live. And I will highlight it in Slido, so it should pop to the top and be highlighted for everybody.
Hi. Wonderful to meet you virtually. So I have a question basically about since you've been a data scientist at Microsoft for so long, how has being a data scientist at Microsoft evolved over the course of your career? And what tips would you have given your younger self now that would have saved you trouble or hindsight is 20-20, essentially?
Yeah. No, that's a great question. Let me spend some time thinking about it. The one thing I will say, it's tougher to be a data scientist, early state data scientist at Microsoft now than it was when I joined. I think data science and that is not just specific to Microsoft. I just think there's been an increased maturity in the expectation out of a data scientist over the last few years, which has developed. That means I think it's tougher to get into Microsoft, but the whole industry is that.
The one tip I would give myself if I joined Microsoft again was move from consulting to product very quickly within a tech company. So when you're in a tech company, we love software. We love systems. We love data products. We like consultants. Don't get me wrong. We like consultants. But what the tech companies are used to are system set up to run processes. They like to bring in consultants from the outside. So for me to succeed, it took me, I think for me and my team, I'd say to succeed, it took us a couple of years to figure out that we may be a talented set of data scientists, but for us to consistently add value to Microsoft, we need software and systems and we need to make it a data product that people can rely on and can plug into other processes. Not like every five or six months, come up with this cool model on how to focus the world, you know, get all the accolades of the great model and then go away. So I think that would be one important tip, Nora, I'd give myself is like, if you're in tech, start moving into product and data product thinking as soon as you can.
Using public data for supply chain planning
Joey, insightful question about Census Bureau data. Would you like to ask yours live?
Thanks. Hey, Sajay. Thanks for spending your time with us. I'm wondering what public data or like Census Bureau data you use in your role, for instance, like planning, supply chain planning for data centers. And what data do you wish you had that doesn't exist?
So full disclaimer, I work for the Census Bureau, but I'm here just in my personal capacity, not representing Census or the government. Just curious. We'd like for private industry to use our data. So I'm curious if there's data you wish you had that didn't exist.
You know what I mean? That's a great question. I am thinking about the wish I had aspect, especially from a demographic standpoint. I'm trying to think to that. Maybe I'll answer the first part. What do we use today? So if you think about, you know, we do a lot of macroeconomic modeling, right? We need to do three to 10 year forecasting for data centers. So we are trying to understand macroeconomic trends within a country and how that will evolve over time. So the first set of metrics you can imagine are the likes of GDP per capita income, tech diffusion, all of those aspects. But the other important aspect that we need to start thinking about also is evolution over time for the type of economy it is. For instance, if you think about a structured economy versus an unstructured economy, right? Like India, for instance, is a lot of unstructured economy. So we need to think about cloud penetration very differently in the unstructured economy and see the trend of how quickly it gets structured and enterprise solutions can go in there. While China is a different economy, it's a very structured economy. And you can, if you have the right tech stack, you can go straight away to the market. And if you have the right conditions for success. So we use a lot of this macroeconomic data to not just forecast demand, but also tell us potentially what kind of strategy we need to deploy with our sales team over here. Do we go, is the market absolutely mature? We just go with our product. And this is in conjunction with them, right? A lot of these things may not be surprised. Some of the especially market maturity isn't a surprise for someone working there, right? Like if you are working in that country, you kind of know it, but helping put it in context of the other countries helps us take learnings ahead, right? Let's say this country was at this market maturity five years ago. What did we do there? What succeeded? What failed? And applying that thinking over here.
In terms of what data I would love to have, I think tech adoption data in general isn't great out there. Like you have so many ways to measure tech adoption. You can think from a consumer standpoint, from a business standpoint, from just the infrastructure standpoint. I think there's a lot of potential for companies to stand up in that space to measure tech adoption because that drives the strategy of a lot of these companies on how they go to market over there. If that helps you.
Tech stack
I have a quick question to sort of interject before we move on to more slide of questions because we usually cover it at some point in the middle and I want to make sure that I ask tech stack. What does the tech stack for your team look like or for you personally? But not every person in like a leadership position gets to code all the time anymore. So it's kind of a combined question. What does that look like for you?
And how do you sort of implement your tools? You know, finally, when I started off coding, it was SAS. Maybe a lot of people, you folks right now don't use SAS, but that was a prevalent technology. Yeah, it's still there, but it isn't as prevalent, I guess, when in the early data science, SAS was quite a big thing. And then I spent a lot of time in R, right? I love R. I'm a big fan. So I try to find creative ways to do things in R, even though they can be done better in Python. I'm like, nah, you know, I can do this better. But I know heart of my heart, there's some processes that Python is just better off. You really know that when you see it. But I still will force myself to do it in R just because I enjoy and I'm so familiar with that language. So I'd say R and Python are my primary go-tos. In terms of the tech stack for my team, we are big. We have a lot of statisticians, they're big on R. So we are a R shop too. Python is,
of course, something you just have to have as a data science shop, right, with all the packages available there. So Python's huge. C Sharp is another engine that we use pretty heavily for the optimization part of algorithms. That does exceedingly well when you really know what kind of algorithm you want and you want to write custom algorithms. I think Python is really good when you want to use off-the-shelf algorithms. But if you have custom algorithm that you know is right for your business and is different from those traditional packages, think about C Sharp engine because of how much flexibility it gives you in setting up the algorithm. And then, of course, you have the traditional data tools that we have. You're sitting on Microsoft Fabric, you have SQL, all of those aspects set up.
Awesome. And what about data storage? I guess you're at Microsoft, it's going to be in Azure, right? Yeah, it's on Azure. It's all in Azure, yes. How do you say Azure? Azure? Azure? I may be wrong. Azure is fine. I feel like we all say it differently. I've always said Microsoft Azure, no one's corrected me, so I'm hoping I'm right.
All right, awesome. And a big question lately has been around notebooks, and there's new notebooks coming on the scene that everybody's really loving. When your teams are working in Python, how are they accessing it? Yeah, so most of them do the Visual Studio VS editor right now, but notebooks especially think... So there are a couple of ways I look at that. Maybe I'm dated, I need to update myself. I like notebooks when you're talking about analysis, which are interspaced with graphs, where you need to refresh, say, a notebook of an analysis frequently enough. I love it. But if I'm moving code to production, I'm like, all right, you know what you're writing, just write it up quickly and move the code to production. So that's more traditional, maybe. I think I need to start thinking about using notebooks more in production workflows too. So we support most notebooks within our team. Most of our developers just end up in Visual Studio, end up coding over there.
Okay, awesome. And with AI agents, it's so easy now to code. Can I say that again and again? As long as you just know your workflow, which is, if you have ever coded, you know workflows and how to code it up. You shouldn't be struggling with code. Right. Awesome. And I was just talking this week with people who had never really deployed anything. Like they never used Posit Connect to deploy something or schedule something. And talking to them about how that is helpful and like all the different ways that you can do it. Does your team deploy things or are you more of a sort of like just ad hoc answer questions type of team? Oh no, we deploy. So one of the, and this is I think, I talked to Rachel about this earlier, is one of the biggest sales for me for Posit, just using Posit Connect on my team, is that my data scientists have now become developers because they can deploy code and I don't need a separate dev team to go productionalize the code.
I need to, of course, you know, all of us need to upscale on that aspect, right? So if you think you're a traditional data scientist, you need to upscale with software engineers on how to modify your code to be production ready. Software engineers need to learn how data scientists think. It's different. Like, you know, it's very different. I think if you have been on both the data science software side, you just intuitively know how differently coding is thought of. But once you get through that hump, your data scientists become developers and that cuts out a lot of dev time. One of the problems we had maybe like four or five years ago when we had separate dev teams was that a data scientist used to develop the models, including me at the time. We used to then send it to the dev team to go to production, which used to take another two to three months. And by that time, the model we wanted had changed.
So that lag time really can kill a model in the business. So that's where I think Posit Connect is a big enabler to convert my data scientists into developers and save me a lot of dev time and value that way. Awesome. It's definitely, it definitely, as a person and developer, a data scientist, it cuts down on the sort of cognitive load of needing to get things developed and out to deployment. So I've been very lucky to be able to have it in the past.
once you get through that hump, your data scientists become developers and that cuts out a lot of dev time.
Audience Q&A: Forecasting AI demand
It's amazing. All right. So I have a question that's, I'm going to put in the chat here from Rohan, and if we could get Rohan unmuted to ask it, I think that would be great.
Hey, Sajay. Great meeting you, Sajay. So in 2022, something called ChatGPT came out, and then now we're in this war where everyone is trying to create the new LLM and foundational model. Of course, foundational models require lots of data centers. So how did the emergence of these technologies change your projections and models for the demand for data centers? And I'm assuming it made it a little harder. How much harder is it, or is it easier?
Great question. I'll tell you, I think data centers were always talked about before AI, but they're talked about way more right now as the enabler in the world of AI, because of a couple of things. One, more power requirements, absolutely. But also that, you know, if you think about large training sites for these LLM models, they need to be contiguous. They need to be in one place so that all the GPUs can learn together, run together, train the model, and they slide down together. So back in, to just take you back in time, like around 2021, we did have something called OpenAI in our systems already, right, training their models. But we had no idea how big it would be. That was a very closely guarded project within Microsoft, including my team didn't know what it was other than being a business project. When we saw AI on what potential, and this, I think that one of the first applications was the image application they had, right, Dolly, which came out, which is publicly available. And I saw that and I'm like, wow, okay, this is game changer, right?
And at the same time, of course, we were trying to forecast AI demand. And it was really difficult. Now imagine three years ago, today, at least, you know, AI agents, you can see the applications. Two years ago, when chat GPT has just come into this world, you're like, the optimist in you is like, oh, the world has changed. The pessimist in you is like, oh, you correct natural language processing. Okay, yes, people have been working on it, you just cracked it. Okay, fine. But that's where we were trying to forecast AI demand.
And we don't have much data to forecast, right? As data scientists, we love history, use models, you learn from history and do it, we don't have history for AI. We spent a lot of time researching the right way to forecast it. And you know, funnily, the way we ended up forecasting it at that time was, we thought about historic analogy, an analogy similar to that, right? So what do you do when you have a new disease, for instance, that you don't have data on, you look for similar diseases in the past, and try and figure out how those diseases spread similar to Spanish flu and COVID. The closest analogy we could come up with at that time, and this is now two to three years ago, right? In hindsight, it may look like, oh, what's the big deal? But today, it is obvious. It was iPhone apps and the App Store. So if you think of iPhone, iPhone, breathtaking, amazing new technology at that time in 2010. But the real economy was not the iPhone, it was the App Store, which added a value to a lot of people's lives, and created that whole economy around it. And that's how we thought about training and inferencing. We said, think of your iPhone as your training model, setting up the baseline for you. But your real value add for customers across the board is going to come from inferencing, and they're going to be applications like the apps, which there are going to be places like the App Store, which will host applications, which will add a lot of value to people.
The closest analogy we could come up with at that time, and this is now two to three years ago, right? In hindsight, it may look like, oh, what's the big deal? But today, it is obvious. It was iPhone apps and the App Store.
And that hypothesis is now panning out through AI agents for us, we can see that come out. And that's what we used to calibrate our AI forecasting model. Now, that's maybe two to three years ago, that was a critical juncture for us trying to figure out what happens. In today's world, one of the biggest complications that we need to deal with is lack of fungibility.
So if you think of supply chains, when you're forecasting supply chain, if you have fungibility, it's amazing, right? Because I can build three factories, say I'm talking about, I can either store diapers, or I can store milk cartons, or everything, it's fungible. Then I'll say, oh, I'll build three warehouses. If the diaper demand goes down, I'll put milk cartons over there. If the milk carton demand goes down, I'll put something else over there. So fungibility is huge. We used to have that fungibility pre-GPUs, where a data center is a data center, it can host most workloads, compute storage can host most things. But with GPUs, GPUs are much more power intensive. And the future generation of GPUs need a different form of cooling and data centers, which is called liquid cooling. That means you suddenly don't have that fungibility. So if I plan a data center for GPUs only, that's all that can go there. And if the GPU demand goes lower, I'm done. I'm sitting on an investment which cannot be monetized. So that is the biggest complexity we are dealing with today. It is, of course, the uncertainty and demand volatility of AI, which inherently will exist in a new technology, but also the lack of fungibility that it leads to in most cloud players. This is the same story for Amazon, Microsoft, Google, or any cloud player.
So I guess one clarifying question, when you talk about fungibility, is it like these GPU data centers can only be used for specific things, or is it something else? Yeah, no, that's a good point. So let's say I'm Microsoft, right? So Microsoft has Azure's traditional compute and storage business and, say, GPU business. Now, in a world where data center space was fungible between both, I just am building data centers, right? I'll combine the demand, combine the volatilities, and thus have reduced lower volatilities of the aggregate, and I plan for it. But now I can't do that, because the kind of data center GPU needs is different from the kind of data center compute and storage needs. So I need to plan them separately.
It's like, I love to give, and this is, by the way, if you are into forecasting and supply chain, this is a huge concept, which I think is underappreciated, is just aggregating demand. Take any opportunity you get to aggregate demand up higher. If you are told to forecast the S&P 500 and you want to have a small cone of uncertainty, do not forecast each of those 500 stocks individually. Focus the S&P 500 in an aggregate, you'll have much less volatility with it. So demand aggregation is a huge construct, which is very helpful in our thinking. Thank you. Awesome. Rohan, great question. Thank you so much.
Audience Q&A: Standing out as a data scientist
And I would love to ask Sam. Sam, if you are available to unmute and ask your question, let's see if we can find Sam. Yeah, I was trying to unmute. Can you guys hear me? Yep. There you are. Thanks. I hope you remember me. Of course. That was a good one. All right. Well, I'm in the job market, and I'm looking for all the new titles that's coming with the Gen AI and I see like AI data scientists. I see Gen AI, MCP data scientists. I didn't think AI developer. How data scientists differentiate themselves and how they can stand out and what makes them unique among this crowd?
That's a good question. And maybe Sam, I'm going to take the opportunity to just fast forward. Imagine a data scientist in 12 months. For a moment, let's work back from, I like to think about like, maybe what's the not-star state, right? In 12 months from now, I'd expect most data scientists to be able to build AI agent apps, right? I wouldn't, it shouldn't be new to them in 12 months. That is a skill set because we have all seen, right, Sam, I think in the hackathon, we saw some of that together. What AI agents can do is fundamentally changing the world and it will change the world. The technology is there right now. As one of the, I think it was, was it Perplexity CEO or Shopify CEO said, you know, tough tasks have become easy. The only truly tough task is what you thought impossible earlier. That's where the world is heading, right, with AI agents. So data scientists 12 months down the line, I'd expect them to know AI agents. I do still think algorithm building for AI is still going to be a separate practice, right? So there will obviously be a core set of engineers focusing on building the next generation of LLMs, adding more features to what AI agent functionalities can do. But AI agent app building should be in the ambit of every data scientist out there.
One of the things, I know it sounds cliche, but one of the things I look for even in a data scientist today, if I were to hire, would be their ability to learn it all rather than know it all. What I know today is irrelevant in six months. I need to know something new. So what I'm looking for is someone who is able to show through their profile and the conversation that they have switched multiple subject matter areas, have learned new things and constantly evolved. My favorite question is your biggest failure. I want to know your biggest failure and what you took away from that. Because that tells me whether this person will be able to learn with me because what I did six years ago in Microsoft, I think I was mentioning the earlier question, it was very different of data scientists' expectations six years ago in Microsoft versus now. And that will change. In a year from now, agentic AI would be like, you know, possibly a basic qualification. We see the basic qualification, preferred qualification. I know today it's in the preferred qualification that will make its way up to the basic qualifications in 12 months. So that's what I'm looking for. I hope I answered your question, Sam. It's more about can this candidate learn more things, learn quickly rather than saying, oh, if I see a candidate, 23-year-old who says, I know everything about agentic AI, I know everything about statistics, machine learning, time series, NLP, blah, blah, blah. Either that person's a genius or they're not true to their resume. So that's how I think about it, if that helps.
What I know today is irrelevant in six months. I need to know something new. So what I'm looking for is someone who is able to show through their profile and the conversation that they have switched multiple subject matter areas, have learned new things and constantly evolved.
Yes, we got a thumbs up from Sam. Thanks for asking that question, Sam.
Big team vs. small team
I love it. Okay. So I put a question in the chat. It's in the Slido, but I think that it's a great one to sort of like jump in if anybody can't answer it because it's about Shiny. But I wanted to go on to an anonymous question, which I think is pretty great too. Which do you prefer, Sajay, a big data team or a small data team? Because I'm team small. I really like being small and intimate.
Small. Why? Great question. So, you know, it's an easy answer in the world of AI agents, but I'm going to go to the answer without AI. AI agents would let you scale, right, your impact.
If you build the right AI agents, you can scale your impact even with a small team. But I'll try and live in a world like a year and a half ago when we didn't know what AI agents and what they could do. I think small teams force you to get creative with your solutions.
One of the biggest problems I have seen in a lot of organizations, especially in tech, tech is guilty of it quite a lot, is they end up amassing huge teams just because that's how they grow. You say, oh, I'll add another team. Oh, I grow again. And now I've added three or four members, three or four members. And then when you break it down and someone new comes into that role and says, oh, what do we truly do? You stack rank what they do. You're like, oh, wow, half these things, I don't think people are using anymore. Why did I have such a big team? And that is the risk of holding a big team. It is very difficult to figure out which of this truly adds value and has that value sustained over time. At a given point of time, it may make sense to add a new team. But over six months, is that team's output still valuable for the company? And does it hold the same value? That's a really difficult question to answer when you have big teams. When you have a small team, you know you have 10 resources. Like I've got to figure out my project plan in a way these 10 resources can deliver to the capabilities. And guess what? If I have more demand than supply, which is supply in my set is like my team, and demand is the projects out there, I get to choose projects that would be of value to a particular person's career. Rather than have to say, oh, yep, because I have a large team, I'm going to take all my projects, prioritize it, and potentially have a super talented data scientist work on something they're not interested in and lose them in a year.
Awesome. Thank you. Thanks for the anonymous question, whoever you are. It was a great question. And we have another anonymous question next. Maybe we can roll right through a couple of these. How do you stay updated with trends in data science and machine learning, Sajay?
And to be even more honest, you know, prior to two or three months ago, I was hyper-focused on work. Just whatever is needed for me and my team to be successful at work, I was hyper-focused on that. And with kids, you feel like you don't have any time. It's in my paternity, which is the last two or three months, I have been using that to bond with my family. Family's great, don't worry about that. But additionally, I have had the opportunity to interact with the community.
So, be it conversations like these, be it hackathons, be it some industry conference, I think I have learned more about what I don't know and what I need to learn in that than I've ever learned. So, you could always try reading up everything out there, but I'm telling you it's impossible. What you'd rather do is be part of communities like these, where there are folks who are going to go read, and you aggregate intelligence. And then figure out, oh, what does the industry really need? Oh, this is interesting. This is something I'm interested in. So, I would suggest think in terms of just being part of more and more communities.
And in every community, look for the smarter community than you. If that community is discussing everything that you already know, you can contribute to that community, but you're not going to learn. So, look for communities where you learn half the things that you learn in the call you don't know about, and you want to go learn. I promise I didn't pay Sajay to say that, but that is exactly what I would tell everybody. Everyone's like, oh Libby, who do I follow? Or what do I read to stay up to date? And I'm like, no, I don't do that. I talk to my friends. That's how I learn about new things, is I code live with other people and learn from them. So, the more you can do that, I feel like the better.
If that community is discussing everything that you already know, you can contribute to that community, but you're not going to learn. So, look for communities where you learn half the things that you learn in the call you don't know about, and you want to go learn.
So, I'm glad that we're in agreement there. I love a community answer. Thank you so much, Anonymous Asker, for that question. It was fantastic. And I have another anonymous question here that's a little bit more technical. If you're up for that one, we can pivot a little bit. How are you handling macroeconomic modeling with recent shocks like COVID-19, like tariff shocks? For anybody who does not do forecasting, shocks are these big disruptive things that sort of have a shockwave pattern or effect when you're looking at demand and forecasting. Are these similar in kind to things that you've had to model in the past, or have you had to carve out some new approaches to handle these types of shocks?
Forecasting through macroeconomic shocks
That's a great, great question. I think I'll maybe give a couple of examples to give you a sense of how we think about it, because I still think the straight, the direct answer to this question is it depends on the goal I'm trying to achieve.
So, for instance, when I'm doing three to 10-year forecasting of a market size, trying to figure out how big this market is going to be in three to 10 years, I don't care about short-term shocks. Tariff shocks may come, they are cycles, they'll go away, there could be a recession that comes in three to 10 years, it'll go away. But I'm trying to understand in three to 10 years what is the market potential of this, so I'm not worried about short-term shocks. So, that is just to give you a perspective of when I almost don't care about the next year's projection, I'm looking at history and saying what is my 10-year out projection? That's all I care about.
But now let me give you an example, a different example of COVID. And this is, I'm going to take you, everyone, back in time, right? Two or three years ago, no, four years ago, oh my god, it's four years ago, it feels like a different world. So, four years ago, this is around 2020, right? The lockdowns have just hit us. No one has any idea what COVID is going to be like.
But we have conversations about an economic depression happening because economic activity came to a standstill with the lockdowns. And we were in charge of thinking about, hey, what will cloud demand look like? And think of Microsoft today, right? You're like, oh yeah, folks are moving to the digital world because they're forced to work from home, so you're getting a rise in demand. But what is the outlook? Remember I told you data center infrastructure is a two to three year game, so you know how to figure out what's going to happen two or three years down the line.
And we did new fundamental research to figure this out, right? We said, okay, let's break the question down. One is, what do pandemics do to pick technology curves? Like, what do they do? So, the only pandemic we knew about was go to Spanish flu. And we studied electricity production. We saw that electricity production, while it was hit in the short term during Spanish flu, in the long term, it actually accelerated. So, pandemics in the short term cause a shock, yes, but in the long term, it actually accelerates. And then we tried to figure out why did it accelerate? Okay, what made electricity production accelerate? And here's an interesting research paper which talks about right after a recession, companies are much more willing to spend money to grow with technology. Because in a recession, sadly, they would have to let people go, and a lot of people, they will be smaller companies. So, when they grow, now they can grow with the latest technology. So, that together, we came up with a hypothesis based on a macroeconomic study that, hey, you know, COVID for cloud, it may actually be a digital acceleration play that is coming through in two to three years from now. And it turned out to be true. It could have been false, but turned out to be true, thankfully.
But if that gives you a sense of how we use macroeconomic data, it's not cookie cutter approach, right? Like, here's the forecast, let's plug that in. It is like, what is my business goal? How can I use macroeconomic data and existing research? And actually, I'm going to stress on that even more, right? Data is one aspect, but there is a lot of existing research out there, which you should go and do a lit review of before you go off to your own project, because a lot of them have been analyzed and you can leverage information from that. That's great advice for more than just this topic. Right. Go do a literature review of whatever problem you're tackling. Thank you so much, Sajay. That was a great answer.
Communities and networking
And the next question really ties into what we were just talking about, which is, are there communities or meetups that you would recommend yourself for networking and learning? Great question, Anonymous. Actually, can we do a poll with your group for that? I'd love to know more communities that I should be part of, because I'm in the zone, as I think Libby mentioned, I'm in the zone of just talking to people. Every time I talk to anyone, whether I'm mentoring them, I'm getting mentored, or just having a casual conversation with colleagues, I'm learning. So I don't have the perfect answer to this, rather than saying, go find and talk to more people.
Awesome. I put something in the chat for everybody under the anonymous queue thing. Are there any communities or meetups you'd recommend for networking and learning? If everybody will hop in that, and maybe create a little thread, and then we can make sure that Sajay gets a copy of our chat.
That'd be awesome. And actually, I see Sam's answer in there. Sam and me both are in Seattle, so we've come across each other. Seattle Data, AI, and Security, I've been part of this community for the last three or four months. It's amazing. It's in-person events for those in the Seattle area. I think there's an event around how to use Microsoft Fabric in your workflow this month. Last month, there was a hackathon. So I can vouch for that community. Great community to be part of. It's a nonprofit. Folks are just here to network and have a conversation.
Awesome. All right. Well, I see a few questions in the chat here from people that I can actually unmute to ask them. They're not anonymous. Rachel, I saw that you asked a question in Slido. Is that something that you would like to ask live, or do you feel like it's been answered? Oh, no, this one hasn't been. Awesome. Sorry if there's any background. I'm just in a conference room today. But I hear a lot about how can people use Quarto and PowerPoint together better or more effectively, and Quarto translating that over to Word docs for people on their team. I was just wondering, how can Posit and Microsoft work together to help make this happen?
That's a great question. We should chat about it. This is where, Rachel, I love to connect with my larger team, the folks who actually get stuff done. I have a big advantage for folks who don't know. I have some amazing data scientists on my team. They do an amazing job. So my job is just directing them. So maybe, Rachel, I don't have the perfect answer, but we should get together with the folks who actually do a lot of the hands-on work on my team and figure a way out. Because the one thing we know with Posit is it's always been a success story for us. So folks are open to collaborating, and let's figure something out over there.
Okay. That's amazing. Yeah. I'm very invested in the answer to this question, too. If anybody knows about the Office-verse packages for R by David Gohel, if you are looking for ways to increase your flexibility around moving something to PowerPoint or something to Word from R, I would highly recommend the Officer package, the Office-verse suite of packages. But they are really for R Markdown to PowerPoint and R Markdown to Word. So the Quarto-specific conversation needs to happen. I'm here for it. Let's do it. Okay. Thank you, Rachel, for that question.
Optimization model constraints
I have a technical question from Sarah. Sarah, if you are available to ask that, that would be great. Let's see if we can find Sarah to unmute her. And I know we only have about 10 minutes left, but hopefully we can get in at least a couple more. I may have tried to unmute the wrong Sarah. So let me just... Oh, Sarah might also have jumped out. Okay. So maybe if you want to ask... Yeah, Sarah, if you happen to be listening later. She asked, do you happen to know any common mistakes that people make when they're defining constraints or objectives in optimization models? So for anybody who hasn't worked in optimization, you define mathematically your constraints and also your objective function, the thing that you're driving towards.
And she also asked, also appreciate any suggested resources or communities for those who are new to optimization models. So I will let you go any way that you would like to go with that. Yeah, I'll take the first one. You know what I've learned about optimization models in my experience?
Technically, the data centers are generally fine. The kind of constraints we specify, they're generally fine. We may make mistakes. We learn from our mistakes by running simulations and figure out what's the right constraint to put in. I don't think the challenge in optimization models is technical in nature. I think the challenge of optimization models is making the business connect. Like some of the things that we may think is a hard constraint isn't really hard constraint, or it could be a hard constraint, but not super valuable.
So the question I always ask myself in the optimization model space is, how can I get the minimal number of business constraints and get that in very cleanly without any guesswork around it? I love to keep my constraints at the minimal. I want more solutions because that gives me a range of possibilities, especially in an MVP to go run against my business stakeholder. So if there's any advice on optimization algorithms I'd give is spend more time identifying from your set of business constraints, not the technical constraints you're going to place on your model, the business constraints, which of them are truly essential while which of them are good to have because that helps you find a much more reasonable, feasible solution. Because if you apply all the business constraints that your business stakeholder just told you, I guarantee you do not have a feasible solution because they are asking for the world and that doesn't exist. So look for the minimal number of true business constraints before you build an optimization model.
Because if you apply all the business constraints that your business stakeholder just told you, I guarantee you do not have a feasible solution because they are asking for the world and that doesn't exist. So look for the minimal number of true business constraints before you build an optimization model.
Awesome. I think that's a great piece of advice. It's like know your data and know the business problem first before everything else. It's going to make all of your other problems and jobs easier. All right. Thank you, Sarah. Sarah is here by the way. Sarah just had a weird Zoom name so we couldn't find her. Thank you so much for the fantastic question. I have another question that I think we can sneak in here and there's an asterisk so I'm going to read it. Samita asked, does your work involve a lot of stakeholder communication? I'm sure it does. If yes, are there any specific data visualization tools that you use while communicating insights? Great question. That's a great question. My business is absolutely
stakeholder facing and that is the reason we succeed I feel because just to give you a sense once my team signals a data center plan it's billions of dollars and gets executed on. It's amazing impact but great responsibility and business communication is absolutely fundamental to us succeeding and the tip I'd give you is like more than the type of visualization like you know the standard visualization charts are great right. If you have two variables do a correlation make it look cool. It generally looks nice. Time series trends always win. Just go with time series trends because it's easy to visualize history and see the forecast and everything is much easier in your world from a cognitive standpoint but the part I'll stress on even more is for a moment be the stakeholder. Be the decision maker and think if I am the decision maker what data points do I need to see to make the decision. Don't make the decision for people right. You know data scientists I've seen operate either in support function where they keep providing a lot of graphs or they go and give the answer straight up here is the right answer.
Human decision making works well when they make it with you. So be the enabler of that right so sit in there and say if I was a decision maker and this is a question my manager had asked me because I had produced a demand forecast for cloud with uncertainty bands and said hey here's my forecast and they did ask this question my manager asked so Sajay what would you do. I'm like what do you mean what would you do. Here's a forecast it's got a 95 percent confidence interval. I checked the back I've done back testing the accuracy looks great the features that are great they're quite stable. He's like no no no I'm asking a simple question what would you do if you were the head of infrastructure with this data point and that made me fundamentally change how I think about it because if I am the infrastructure plan head what do you need. I need a forecast.
I potentially know need to know sensitivities of my model where I may be over forecasting under forecasting based on the assumptions I made and I need to know cost of over and under forecasting because what if I'm higher what if I'm lower because no one knows the answer.
So an easier way to think about it is put yourself in that person's shoe and say if I were making the decision what are the data points I need in front of me to know no decisions are right but this is the best decision we can make. Yeah I want to ask you a little career advice question in here.
the best decision we can make.
Building strategic awareness as an IC
How do you get to know your managers your stakeholders your business leaders so that you understand what they're worrying about what they're sort of like optimistic pessimistic view of the model outcome is like how would somebody in an IC role sort of gain that subject matter expertise that's that level above them that's more strategic than tactical.
No that's a great point I think strategic is the key word what I've learned especially through my one-on-ones my earlier one-on-ones early in my career used to be very much like hey here is what I have done last week this what I plan to do next week what do you think it used to get very tactical about that problem statement and narrow down my manager's focus let me just focus on the manager maybe right it used to narrow down their focus what I've learned over time is to ask a different question the question I ask is hey what are the top three biggest things that's worrying you like what is keeping you up right and then you'll hear much more strategic problems you don't hear tactical problems you hear like I think we get lost in the data or we are not analyzing this big business area because we don't have bandwidth here is a trend that is coming but we are not catching it you hear the strategic problem that people have and you'll be surprised how open people are to have that conversation if you untie it from a deliverable due this week or next week you talk about a deliverable due next week this week a project plan you're going to get rabbit hold into that question but talk about bigger picture saying hey what are the bigger things that you're worrying about tell me the top three things that your leader is worrying about you will get a human sense of the things that worry you it's similar to the conversation you'll have with your partner right if you talk about an issue that say the kid is not eating blah blah you're gonna get rabbit hold into that but just take time and say hey what's going on what do you think what is keeping you up right now and you'll hear other things you're like hey work is increasing and I'm worried of the kid growing up at the same time and the work increasing how do I handle both so you want to get out of the tactical problems and say get the high level picture so that you focus on the right things in your analysis.
yes I love it this is similarly how I think about financial decisions I think Morgan Housel is the one who who suggested like make the financial decision whether that's investing or or whatever that helps you sleep at night like that might not always be the thing that's the optimized choice for you but it's gonna be probably the best choice long term um and so that kind of falls in line with this like what what's gonna keep you up at night if you if it comes around actually actually let me one more point this is something because I got this advice from a mentor not so long ago like when I think about I was asking a question what is how do I define success for myself in a role it's a very difficult thing right you can always think in terms of how work measures you but how do I think about it for myself and she said something very interesting she said the way she thinks about success is can I operate one level higher and one level lower in my role seamlessly if I am able to do that without getting stressed without having to stretch that tells me that I'm doing really well at that role because I can go help my team members and help my boss as and when needed that means I'm ready to grow if I can help my boss and I can help my team members if I have to go one level lower and do the detailed work with them that was one of the best advice I've ever received on how to think about success in a role.
the way she thinks about success is can I operate one level higher and one level lower in my role seamlessly if I am able to do that without getting stressed without having to stretch that tells me that I'm doing really well at that role.
that's amazing I feel like I didn't even ask you the right question but you gave me the right answer I don't know if I've always been able to answer that question I guess the is the follow-up advice like if you can't go figure out how to fill those gaps yes absolutely I think for instance I have gaps in going both one level higher and one level lower I don't think I can seamlessly do my boss's job right now I have things to learn similarly I have some areas of detail to learn for my managers report to me so I can't just replace them say they are on a vacation I don't think I'll be able to completely do justice to what they do so that is and now I know the improvement areas for me to get there.
Closing and resources
amazing well we have two minutes left so I am not going to go any further Sajay thank you so much for hanging out with us you have been a fantastic guest I am already getting like pings on the side that are like Sajay's awesome he's so amazing he's such a great speaker so I wanted to put your LinkedIn profile in the chat I will do that for everybody let me see and everybody if you don't know there is a little three dots the top of your chat you can click that and then save chat and then all of these resources that everybody has shared will be saved and you can have them so everybody go connect with Sajay say hello to him go find other places where he's he's talking Sajay I know you've done like panel discussions and stuff in other places oh yes I'm sure there are lots of places to find you. Can I give a call out to ADP list if some of it would help to have one-on-one chat from a mentoring or just a connect perspective where mentoring has a word just connecting with me I'm on ADP list where I'm talking with folks so feel free to schedule time with me over there all right I just put a link in the chat to adplist.org it is where you can meet with mentors and next week I would like everybody to come hang out with Pallas Horwitz analytics consultant and professional development coach if you would like to have more conversations about professional development in the management space or in the data science space Pallas is a great resource for that so hang out with us next week to meet her and ask her all kinds of questions thank you Sajay thank you thank you everyone I appreciate you spending your time over here thank you we appreciate you I'll hang out for just a minute more so that if you would like to save the chat file you can do so thanks everybody have a great rest of the day can't wait to see you next week bye everybody.
