This episode features an interview with Sunny Bedi, CIO of Snowflake. Sunny has previously held senior positions at Deloitte, VMware, and NVIDIA. In this episode Sunny talks about how Snowflake is a data-driven company, data security in the cloud, how to use AI to minimize data threats, and much more.
This episode features an interview with Sunny Bedi, CIO of Snowflake. Sunny has previously held senior positions at Deloitte, VMware, and NVIDIA.
In this episode Sunny talks about how Snowflake is a data-driven company, data security in the cloud, how to use AI to minimize data threats, and much more.
--------
How you approach data will define what’s possible for your organization. Data engineers, data scientists, application developers, and a host of other data professionals who depend on the Snowflake Data Cloud continue to thrive thanks to a decade of technology breakthroughs. But that journey is only the beginning.
Attend Snowflake Summit 2023 in Las Vegas June 26-29 to learn how to access, build, and monetize data, tools, models, and applications in ways that were previously unimaginable. Enable seamless alignment and collaboration across these crucial functions in the Data Cloud to transform nearly every aspect of your organization.
Learn more and register at www.snowflake.com/summit
Steve Hamm: [00:00:00] So you're a very busy guy. You have three roles, three leadership roles there at snowflake. Chief information, officer chief data officer, and also you head up corporate security. I think it would be really good for the listener. If you could start by talking a little bit about each of those roles and how you put your own stamp on them.
Sunny Bedi: [00:00:21] I definitely have all the three roles that you've mentioned. So let's maybe start with the CIO role. so as you know, we're, growing up business, or, have scaling, challenges and we're getting into new markets.
So we need business process systems and automation that. Needs to be done in a very graceful manner. And so we have a lot of initiatives in the it space, to partner with the business, whether it's in sales, finance, marketing, support, organization, engineering organization, and help build, the processes.
And the infrastructure that's needed to scale very gracefully. so that's probably the most overarching charter for it.
on the uh, chief data officer, portfolio. I like to emphasize that, Using our product internally is top of the initiatives that we have. So snowflake on snowflake is a very important initiative for us internally. And the way we're partnering with our product management and engineering team is we want to be customer zero.
as soon as they are getting ready to go to market, months before that, we want to basically be the first customer to use it. And give them as much feedback as possible, so that they can, really Polish the product for prime time and go to market with a very good experience for the customers.
And then finally the security, as we. led up to our IPO. There was a lot of, compliance initiatives that have been kicked on for the company. And, um, you know, some of them are regulatory requirements and to new markets, we're trying to get into.
And some of them are also requirements that as you go become a publicly traded company, you have to fulfill those obligation. So that's how, the three areas come together. and, um, we can go into more detail as you wish to.
Steve Hamm: [00:02:28] I did actually want to probe a little bit into the chief data officer role and that's because a lot of organizations claim to be data-driven, it's one of those catchphrases of the day. But I'm not sure that they are necessarily. so how do you define the term and what are the most powerful ways that snowflake uses data?
Sunny Bedi: [00:02:51] So I think our, first principles in the company are that if we measure the right setup, thanks. we believe. It will drive good business outcomes. that's fundamentally, our first principle thought process in every organization and the company and being a data platform company, we have to really, live up to that, promise.
So I can give you some examples in the it organization on how we use it. in essence, for it organization service now platform is a very important platform on how we actually do all our work. how the tickets come in, how we communicate with our end users.
And how we actually do change management, problem management, every aspect of what an it life cycle workflow looks like is all managed in our service. Now platform that we rely on, it's a SAS based application.
We have integrated. All of the data that sits in service now into our own internal instance of snowflake, which we call snow house. All that data gets ingested in it. We start in it. Our first meeting on Monday is a meeting call, operations metrics review. And what we do is for all the verticals of it.
We spend a good solid one to one and a half hour in inspecting all the key metrics that we really care about. And we start the week like that. And, it really gives us a good pulse of how we're performing. and that meeting is, not just for me and my staff is pretty much for the whole it organization. Everybody's invited, we have a rotating agenda every Monday, and we go through, all the different pillars of it, whether it's. security, whether it's, availability of a service, whether it's, reliability of our service or it's, employee user experience.
So we measure all of those dimensions and, uh, it's all inside snowflake and it's visible to all of it employees and they know that they're being measured, and there's full amount of transparency. And visibility for all employees that manage those workloads.
Steve Hamm: [00:05:13] So it's not just the manager looking at the data and monitoring it's it goes out to each individual. So they know about their own performance and how they fit in.
Sunny Bedi: [00:05:22] That's right. That's right.
Steve Hamm: [00:05:23] Now I wanted to go back into security a little bit because you know, we're talking about data and, going back in history, there was a lot of concern among enterprises and reluctance to move a lot of their data and their applications to the cloud, because concerned about, about data security.
That's been overcome at this point, I believe. But if you could kind of walk us through, how secure is the data in the cloud these days and compared to on-prem data?
Sunny Bedi: [00:05:58] I’ll give you our example for corporate side. so we are using the CIS framework. Uh, CIS framework is very similar to nest or ISO. I'm sure you're familiar with those frames. and for us in snowflake, we don't have any workloads that are inside.
A data center or anything, on prem, every workload that we have is in the cloud. so it's a, it's an incredible opportunity and an, and, situation to be in, because you're not really worrying about. the constant operational headaches associated with an on prem environment, uh, would be partner very closely with the SAS providers in ensuring and validating how they're keeping their security compliance posture up to date.
And, we build automation around things that is our responsibility from a patching perspective. and then we're using our own platform to ingest all the security data that’s, absolutely needed for, monitoring, measuring and. Meeting the CIS control framework that we have implemented and the most beautiful thing about this thing, Steve, is that if we do see anything that is suspicious, we have built so much automation and with service now, platform that it itriggers and opens up a ticket for the security professionals, the it professionals to take action.
So it's really event-based. remediation, that's taking place, not really after the fact, majority of the times when you are in a non prime environment, it's not that your it or security guys don't want to. I act on it, but the problem is the signal is late. The signal, it gets to the person who needs to act on it in a minute on timely fashion with our platform, we're ingesting all this data.
We're getting informed about it. We have written automation to trigger an action that needs to take place and it's happening, you know, in instant time.
Next path of this is to actually build more automation where it's, it's being done. through AI and ML where, we can even have actions taken on its own where we're not even dependent upon the ticket to be initiated.
It's done automatically if that event were to happen.
Steve Hamm: [00:08:30] Hey, does your AI enable you to do really weak signal detection and things like that?
Sunny Bedi: [00:08:36] That's right. That's right.
Steve Hamm: [00:08:39] so it's not just automating, but it's also really finding patterns that may be a human would not. Necessarily connect with the data breach or some kind of a problem.
Sunny Bedi: [00:08:50] Yeah. I mean, a good example of that is insider threat detection is a very big use case right now. Right? Where. let's just use an example if you have an employee or a contractor, that, is not intentionally trying to do a bad thing, but let's assume for a sake that
Were there last week and they were trying to download it a lot of data from one of the key applications on key workloads, what we want, no, that pattern, you know, if that person was only downloading very simple sets of data. And for some reason we see a huge spike in data that's being downloaded if it's a bad actor, we want to know that real time.
And we want to know that. Anomaly that you're talking about real time and prevent, protect the company for not getting that data in, into any, wrong person's hands. So there's a lot of AI use cases around inside of them thread detection that we're automating as well.
Steve Hamm: [00:09:45] you've been at snowflake since January and obviously a lot has changed. we've had the global crisis, the economic crisis, the, the IPO, all sorts of, uncertainties in the environment. So you when you came in to kind of rethink and build a new it organization.
But at the same time you had all these external issues. So if you would just kind of walk through how you assess things, how you reorganized, how you set things on a different path, and then also how you responded to all of these incredible uncertainties around the company.
Sunny Bedi: [00:10:23] Sure. I started like last week of January, and obviously, you know, the forest month when you join a new company and you're adjusting, you're acclimating yourself.
So I was in more of a fact finding mission the first month, meeting, pretty much every department meeting my team. meeting all the, yeah. and more in a learning more right. Trying to understand what the current landscape is, where there are challenges, where there needs to be focus.
and we started to get information about colon COVID that's really, growing in this country. so the first thing I did was I basically told the it and the security team. That, let's try to not come to the office for an entire day, and that's not informed anybody about that. So pretty much let's do a remote work from home day, the entire team.
Like I don't want anybody in ITO security show up and let's see how we actually service the company. we did that end of February timeframe. That's when kind of code was picking up a little bit in New York and other areas, and hadn't trickled in itself into California, but we had employees in New York.
So we started to, really practice BCP business, continued planning in a way. And we learned a lot from that exercise and sure enough, in two weeks down the road, March 10th, we pretty much do decided as a company we're going to work from home. And, uh, fortunately, because as I mentioned earlier, we don't have anything on prem.
everything is in the cloud. We had to tweak some business processes, but overall, out of intensity, in servicing our employees from a remote perspective, really increased. And, I'm extremely proud of the team on how we have actually, adjusted and fine tune the processes that we needed.
Steve Hamm: [00:12:27] Snowflake is one of those cloud native companies and the Valley is full of them, but the world is full of companies that, were on-prem first and probably still are a lot on prem.
Do you see a big trend shifting here? Do you see a lot of companies moving more rapidly to the cloud?
Sunny Bedi: [00:12:48] I think everybody is going to move towards the cloud. That's a given. that transition is going to happen in some companies, it was not as much of a priority in the past, but with the COVID the situation that we have gone through, I think it becomes a forcing function and, I know so many of my peers who have a lot of workloads on prem, they're all trying to accelerate in the Valley.
They're all trying to move things. So the cloud and I would say that in an prem environment, you are wasting 30 to 40% of your bandwidth, and really operational challenges that you. You don't have to deal with in a cloud first company. And, um, I mean, that's kind of a rough estimate of a number.
It could be even higher in some companies, uh, but all of a sudden, you know, the it professionals are really trying to accelerate the transformation that the companies need from the it organization, because they are dealing with the Aldo on-prem operational challenges that come with it.
Steve Hamm: Since you came on board, what are some of the most important initiatives in it that you've already accomplished?
Sunny Bedi: [00:13:55] we implemented a new billing system. That'd be went live in June timeframe and that's a deep collaboration with a finance organization, our product management team and our, engineering team along with it. And that is the core of how we actually do metering, with our end customers and how that actually.
cause we're a consumption based model company. everything around how we actually build customers becomes extremely critical. It has to be built with high degree of automation, high degree of, integration that it integrates with, our financial system, the old system that we had, had scaling challenges.
And so we worked with our engineering team to build this capability, in house. And now it integrates very seamlessly with Workday. So that's a major accomplishment for it. It's also part of our Sox compliance. It's fully bolted with compliance because it's a very critical revenue generating system..
Steve Hamm: [00:15:05] so that sounds like a, like a really important initiative. when you look forward, here we are, you've gone through the IPO, you know the company's growing up fast. What are the most important initiatives that you have before you
Sunny Bedi: [00:15:20] So we have a lot of initiatives that are in flight right now. I'll give you some flavor or a couple of them. One is, around. having the ability to do business in all parts of the world, it requires us to have multicurrency capability implemented. and so we're actively working on that, to improving our compliance posture and making the collaboration work between legal.
Sales ops and finance because we do a lot of contracts with a lot of customers. we need like a better workflow. And so we're implementing CLM contract life cycle management system, both on the sell side, as well as on the buy side. And making sure that any new customer that'd be on board, it's onboarded with higher degree of compliance and speed matters in these things, right?
Onboarding a new customer. If it takes multiple days to get contracts, a red line and back and forth, that happens. you want all of that in a system. You want that to be in a very seamless manner. In a highly collaborative manner. So a major port portion of that is very manual today and we're trying to automate the heck out of it.
you know, snowflake on snowflake is a continuous journey where. Putting a lot of workloads, in that. and then box is another major initiative for us for this next 12 months. we have over, 30, some applications that make up the socks for. And we have to be highly compliant on all of them because they touch financials.
They touch where Avenue. a big portion of that needs to be, tested and get ready for prime time.
Steve Hamm: [00:17:09] So you've talked about snowflake on stuff like for your it applications. What are the most compelling applications that other departments are using?
Sunny Bedi: [00:17:21] the sales team, is using, they're also taking a lot of the data that sits in, Salesforce, customer account data and, in some cases, the customers have not become customers yet, but they are leads and they want to basically do account scoring or they want to do account enrichment.
They want to know where there is, renewal opportunities. here's all kinds of, opportunities to look at how to grow the business, both onboarding new yeah. Customers as well as taking the existing customers and figuring out where there's opportunity to scale that. All of that data sits either in Salesforce or some of the complimentary systems that make up that data.
So they're ingesting all that data insights, snowflake. And obviously they're trying to use snowflake to do a lot of the predictive forecasting, and analytical work that needs to go with it. in trying to gain more market share, gain more share with the customer, or onboarding new customers that are not onboard yet.
Steve Hamm: [00:18:31] that's interesting because he got Salesforce and he got Tableau, one of their companies now, so are they, are the salespeople using Tableau is the front end into this data.
Sunny Bedi: [00:18:42] So Tableau is used at snowflake more on the visualization side and, we're actively, rolling that out as we speak, to a much broader set of end users. in the company who, have the same type of use cases. so yeah, it's more used in snowflake for the visualization side of the house,
Steve Hamm: [00:19:05] You talked before about how. snowflake is kind of customer zero for snowflake. That's a must be very exciting, but it also involves extra risk and extra headaches potentially. How has that gone with you? I mean, is it, is it smooth sailing or do you, really have to do some, important feedback to the product groups to get things right.
Sunny Bedi: [00:19:30] I would characterize that we're still early in that journey. I wouldn't say that we have matured that process to the T. we have some really great initiatives that we working with, our product team, our engineering team, where, we definitely are customer zero. And, the challenge of that is, in some cases, you know, if some of the end users would argue, well, I don't really want to test everything.
Until it's fully ready. but we don't see it that way. We're turning around to change the mindset that we have to evangelize and operationalize that internally, even if all the features are not there yet, that gives the feedback to the engineering and the product team to develop a roadmap with us on what capabilities are not there in the day, zero milestone, but will come in coming months or coming quarters.
And it's systematically creates a roadmap planning with the engineering and product team. And it keeps a healthy tension between both teams. I think it's super important to do that right. And, we have some really exciting, new initiatives that we're working with them to really test that model out.
It's been a couple of quarters now that we're doing it and we're maturing that, and that will continue to mature as we grow as a company.
Steve Hamm: [00:20:49] What does snowflake not do that you'd like to see it to do? I mean, you're giving this really elemental feedback. What are you telling these people, you know from the product divisions that you want?
Sunny Bedi: [00:21:02] So there's a, there are different verticals, but then within the product and for each of those verticals, we are giving use cases, which we are turning on, that we see some challenges and we continuously give that feedback. A really good example of that is data sharing. If I can get all of my 200 plus apps vacations that I have, You know that I have in production all to be on data sharing.
Boy, I would love that because then I would never have to worry about ingesting any of that data. And some of the operational challenges that do exist with, uploading all that data into my snowflake instance. So we're working very closely with the product guys to prioritize where we can partner with our SAS companies that we use and the providers too.
Have either an automated connector in place and or data sharing in place so that we have the whole experience to be absolutely seamless. And that reduces a lot of the operational challenges and it basically is the data cloud story. so it's a roadmap, we know it's going to take time to get everybody there.
it's ton of momentum in the marketplace, that the product team is driving. and, if I could get all of them to be done with that, like today, I would love it, but it's a journey that takes it does take time to get everybody on the same page.
Steve Hamm: [00:22:30] I understand that one of the big trends in the industry and with snowflake itself is that a large number of SAS companies and enterprises as well are building their applications on top of snowflake. And obviously you're doing that too internally. So if you would talk about the advantages of that approach, I think that'd be really useful to people.
Sunny Bedi: [00:22:51] snowflake automates so much of the operational overhead for software developers so that they can focus on their product. They don't have to deal with, the operational overhead that you would be dependent on. so that would be the first second. It would be the snowflakes elasticity allows SAS companies to deliver.
Very fast performance and speed, to the end users. and you only pay for what you use. So what a beautiful model. And then finally the third would be with our data sharing and our data marketplace. companies can share data with their end customers.
In a really seamless fashion, which is a much better solution for everyone. so I would say those would be the three things that I would like to highlight as the approach.
Steve Hamm: [00:23:39] Earlier in the conversation, you talked briefly about machine learning and some other AI techniques that you're using. And I wanted to ask you to expand on that because there's so much AI going on these days. And, um, so if you could just kind of walk through some of the uses, the key uses of AI and how it makes a big difference for you.
Sunny Bedi: [00:24:02] So, for AI use cases,you know, it's all about anomaly detection, prevention and action around that. so the two use cases that I like to highlight one is, the insider threat detection, we're doing a lot of, work on that front to make sure that we are monitoring any anomaly detection through AI workloads.
And then have the ability to prevent that real time. Stop that real time if something bad is happening. So anything malicious that's happening and, uh, the other one that we're trying to work heavily on that in the next couple coming months is, Pen testing.
So pen testing, it's like an offensive security posture where you get some hackers to come in and you actually try to break the system in a fashion that you know, will expose any weakness in your security posture.
And we think there's a lot of opportunity where Rican actually AI, M L a too, do that, not through, humans, but actually do that through an AI mechanism. And so, we almost want to do that repeatedly. We almost want to do that across multiple workloads across our ecosystem. so we're looking at, investing in that part of AI use cases that we haven't turned on yet.
Steve Hamm: [00:25:30] Now when you look ahead, five years at the data cloud, where do you think this is going? What new technologies and capabilities do you expect to see and how will it change the landscape for your customers?
Businesses, government education, whomever.
Sunny Bedi: [00:25:46] Yeah. So I would say that I would like to, okay. Answer that question in three pieces. the first would be that I think That will be an extreme migration to the cloud by the whole industry. I think covert as been the forcing function now to accelerate that cloud migration even faster than what we had all anticipated, and second is I would say there will be an acceleration of SAS providers and companies doing their own AI ML, if you think about how many workloads do you actually use AIML today?
No, that number would be in very small in majority of the companies. in fact, all the companies, I would say that looking five years forward, there will be an increased amount of workloads that will be powered through AI and ML, where you're not dependent upon humans to do that, you know, in a sensible manner.
and then finally, third, I would say is that. To make those first and second things enable, organizations are gonna have to, rethink of the processes, the tools, the personnel, the employees that they have to have. The expertise that's needed, that they can handle these two major transitions that are in front of us.
I would say that's how I would characterize the next five years in this space.
Steve Hamm: [00:27:18] so there's going to be a lot of upgrading of skills. In addition to the upgrading of technologies.
Sunny Bedi: [00:27:23] That's right. And if you don’t invest in folks who can actually, do the AI ML use cases internally in it. you're not going to be able to Polish up those use cases.
Steve Hamm: [00:27:34] Well, Sonny, it's been great speaking to you today and it's just amazing to see all the things that you've done in it. And data management and security there at snowflake in these few months.
so thanks again so much for your time today.
Sunny Bedi: [00:27:50] Thank you, Steve. Appreciate it.