To keep current on tax issues, TEI members always like to hear directly from top Internal Revenue Service officials. So, we were thrilled when Kaschit Pandya, deputy chief information officer at the IRS, accepted our invitation to be interviewed by Brian Kaufman, vice president and tax counsel at Capital One Financial Corporation, at the May 2022 Tax Technology Seminar in San Francisco. Here’s an edited transcript of their conversation.
Brian Kaufman: Kaschit Pandya has been with the IRS for nine years and has a passion for technology. He and his team have done tremendous things at the IRS in the last few years, especially during the pandemic. We’re really excited to have him here today. This is the first time we’ve had an IRS speaker at this conference, and Kaschit was gracious enough to travel from Washington, D.C., to join us. It’s funny—as we sat through the presentations this morning, I couldn’t help but think that the IRS is actually on a very similar journey to what we as taxpayers are. They are dealing with the same type of technology challenges, some even much older and harder to solve, and they’re working along the same track. So, with that, TEI has partnered with Kaschit and his leadership team to form a technology working group that we’re very excited about [and] which we started last year. That’s how we’ve built a relationship with him and some other IRS tech leaders, with the goal of working together as part of a coordinated effort on technology, where we as taxpayers can help them understand where there are roadblocks and opportunities for improvement and where he and his team can help us. It’s a really collaborative process, and we’ve just gotten started over the last year or so during the pandemic. We’re excited to expand and get more of you involved and interacting with the IRS. As technology improves, I foresee lots of opportunities for interaction. With that, I am going to turn it over to Kaschit, just to give some background on his role and how the IRS is working through modernization.
Kaschit Pandya: Thank you, Brian. Good afternoon, everyone. It’s been a fun session this morning, just introducing myself as the IRS. In addition to that, I am now interrupting your lunch. So, you know, that makes me doubly popular. My name is Kaschit Pandya. I’m the deputy chief information officer of the IRS, and I’ve been with the agency for nine years. I came from the private world. I worked at various large organizations—national, international—all in information technology, and it’s been quite a journey for us at the IRS over the last many years. It’s been an incredible ride for me, experiencing the government side after having worked in private industry for so long.
As Brian mentioned, we’ve been working collaboratively over the last year and a half or so, right around when the pandemic started, and the intent of the conversations that we’ve had—and I really do wish that, after this conversation today, many more of you are able to join those conversations. At the IRS, certainly as a tax revenue agency, we are heavily focused on how, from a technology perspective, we can improve the experience of everyone. While a lot of the focus tends to be on the individual side—because, again, the individual side is larger than the corporate side—my perspective as a chief information officer is to determine how technology can best enhance and improve the experience across the board, individually and on the corporate side.
The IRS has taken on the modernization efforts starting in 2019. We initiated a multiphase modernization effort. We called it “Mod Plan 1.0.” Naturally, right? It’s a tech term, and we love our tech terms. Mod Plan 1.0 was broken up into two phases, three years each—six years total—and the focus of it was across four different pillars: taxpayer experience, core tax services, [and] two pillars focused on IRS operations and cybersecurity. The IRS operations and cybersecurity are inwardly focused, while the taxpayer experience and the core taxpayer services are outwardly focused. We want to make sure that we are able to provide the best customer service and customer experience to all of you—corporate, individual, regardless. But we often realized we were not looking inwardly enough from a modernization perspective. We want to make sure, of course, that every service that we can modernize from a taxpayer experience perspective is available and accessible to all of you, but we also knew that we had to look inwards to improve our operations. And that’s why the pillars are broken up into outside and internal facing. Cybersecurity oversees all of that, because security is of paramount importance to us. But modernization is focused on the outside as much as it is on the inside. So, Brian and I, we’re going to chat a bit more about what we’re going to do in the modernization space, externally facing and internally facing, but really focused on the full spectrum, outside and in.
Kaufman: I think a lot of times you don’t hear about some of the really good work done in the corporate space. For example, during the pandemic, Kaschit and his partners and I were able to push forward first an upgrade to government Zoom. And now it’s to the point where I think almost every forward-facing agent in the IRS has access to Microsoft Teams, [allowing them] to go to a video environment, which is a huge benefit, particularly now when we’re in an environment where we’re just starting to get back together in person. A couple other things, the IRS rolled out a secure messaging site to allow the transfer of large documents. That is something our TEI group had encouraged, and Kaschit and his team worked really quickly to do that, as well as to make accessible to their teams some of the virtual reading room technology out there and to white-list a variety of programs to help exchange information, particularly in transfer pricing and R&E audits. And that’s just the start. I wanted to ask you, What are some of the challenges? Obviously, the taxpayers have a different type of budgeting approval process, particularly those who work in agile [practices] and can be a lot more flexible with spending, with hiring, with developing and failing and then pivoting, which obviously is much harder in a government environment.
Pandya: Absolutely. The biggest challenge that should be no surprise to anyone, really, is funding. Consistent, multiyear funding is a challenge for us as an IT organization—and just to provide a little bit of a scale of what the IRS IT organization is, IRS IT full-time employees number about 7,200. Once we add the contractors, it’s over 10,000. That’s just the IT organization. Our annual budget is about $3 billion. It’s a large organization with a huge scale and a huge responsibility. But that consistent funding remains a challenge. Of the $3 billion, approximately eighty percent of that goes toward O&M—operations and maintenance. So, we are limited in how much we can devote or allocate toward any of the modernization efforts. All of the various technologies that we want to pursue or need to pursue to improve the services, the capabilities that we want to provide, are always a challenge because we are focused on making sure that we are able to provide what we already have. All of the legislation that comes our way, whether it’s the Taxpayer First Act or IIJA—the Infrastructure Investment and Jobs Act—the Build Back Better—that doesn’t all come with funding. So, we are challenged by the fact that we have to take or allocate what we have been given as an annual budget and determine where and how best to provide on the modernization efforts. So, funding is the biggest challenge for us.
The Pandemic
Kaufman: So, for example, concerning the [stimulus] checks that came out during the pandemic, building a process for that came through your organization.
Pandya: Absolutely. In 2020 and 2021, there were three economic impact payments—EIP—better known as stimulus payments. We were responsible, I was responsible, for helping to get those stimulus payments out. Just for comparison, in 2018, when there was a financial crisis, the first stimulus payment that went out took seventy-five days. In 2020, after the CARES Act passed on March 27, we had the first stimulus payment out by April 13. That was two weeks between when the CARES Act passed and when the first stimulus payment went out. The second stimulus payment was determined, or at least determined to be provided, on December 27. We made the payments on December 29—two days. The third stimulus payment came out on March 16 [2021]. We turned that stimulus payment around within twenty-four hours. So, we are constantly looking for opportunities to modernize, but really how to make sure that the taxpayer experience and the impact on the economy, on the American taxpayer, the citizens, is the best it can be. Amongst the three stimulus payments, and then, of course, the child tax credit came out last year, we have pushed out $1.5 trillion into the economy, all under COVID, all under remote access, all under limited engagement, at least from [an in-person] perspective. We are constantly looking for opportunities to modernize and improve our services. And we have done some remarkable things over the course of the last two years. I’ll keep coming back to the challenges, of course, the challenges being [that] we are not always funded for what we want to do.
Tax Technologists
Kaufman: With that, Kaschit, can you talk about the workforce? Obviously, the people in this room are tax technologists, which is a relatively modern term. From our conversations and working with you, it seems like the IRS is trying to develop the same cadre of technology professionals who have expertise but can make their way through the tax code as well.
Pandya: Absolutely. It’s a change in the paradigm of how we provide technology services. Previously, the technology organization was heavily reliant on our business part of the organization to tell us what is it that we need to focus on and how best it is that we can deliver something. The technology was provided, and it was built around what those requirements were. But from a technology side, we were, I would say, almost the caboose of the [train]. If you think of a train model, the engine up front was the business and the caboose being IT. We are changing that dynamic. We are now not only incorporating but sitting at the same table for the conversations of what the technology needs to be to deliver on services for taxpayers. I mentioned the workforce—7,200 to 10,000-plus with the contractors; we are actively hiring and have onboarded multiple technologists with the focus on tax technology. So, we are talking about and thinking about, and actively planning and executing against, a future state of what a tax organization needs to be from a technology perspective, not simply waiting to hear from our business partners what that future state will be. So, how we are approaching the future state—by bringing on individuals who are technical but fully aware of the tax impact itself—is very, very different [from] how it had been in the past.
Kaufman: Knowing the budget limitations and, even more so, the approval limitations, how do you balance between the use of employees and contractors? How does that decision-making work?
Pandya: That’s a great question. We have a very skilled workforce. We are faced with the same challenges, or similar challenges, that we are likely hearing about and seeing in the industry. Our attrition rate is about four to five percent annually, and we are trying to ensure that we are focused on not just backfilling the individuals who have left but also bringing on individuals who have the right skill set for where we are headed in the future. We are heavy mainframe, heavy COBOL, heavy ALC, and this is what [the press] tends to refer to; there isn’t a headline I’ve seen that doesn’t somehow reference Kennedy-era technology. I invite all of you to join us at some point and see our data centers. We are very, very modern. The language, however—we have built applications using COBOL. We have built applications using ALC. Rather than backfilling the individual who has left with that ALC and COBOL knowledge, as we modernize to use your Pythons, your Java, we are bringing in individuals with the skill set who are familiar with and know multiple modern languages.
Looking to the Future
Kaufman: So, where are you headed in the future? What is that vision? Obviously, you have some ancient mainframe systems where data is stored, but the world is going toward the cloud, and I suspect the IRS is trying to get there, too.
Pandya: Absolutely. Just a quick correction: It is not an ancient mainframe. The mainframe is as modern as last year. It’s a brand-new system that we built. The infrastructure, the hardware itself, like I said, is absolutely modern and current. Where we’re headed, however, to [address] Brian’s point, is to leverage more and more of the cloud capabilities. We are actively and aggressively pursuing cloud capabilities. In fact, just this year, I’m responsible for—and we’ve successfully executed on—starting with some of the applications that we’re going to move to the cloud. We started with what I refer to as commodity services: your email, your Teams, your Project, your OneDrive, all of that’s being moved to the cloud. We are actively building out the hubs in Amazon as well as Microsoft Azure so that we can start migrating the applications. We have 700-plus applications. We are now actively working on migrating between, I would say, forty and sixty applications between this year and fiscal ’23. Data is another critical area for us. As all of you can likely imagine, we have voluminous data. And when I talk about the amount, I’m talking about petabytes. Right now, we are looking at and probably have about twenty to thirty petabytes of data. Security is absolutely paramount. We are being absolutely very, very critical of what we migrate to the cloud, because we want to ensure that it is safe. We are fully FedRAMP certified, and when we migrate, we are ensuring that we are moving to an application and a data structure that is fully FedRAMP certified. I wanted to put a plug in: We went live with our very first data in the cloud just three days ago. Of the twenty to thirty petabytes of data, a lot of it is unstructured. We are developing and deploying an enterprise data platform that helps create the structure, create the right amount of data lakes and data stores. I said three days ago; today is May 2, so probably four days ago, excuse me. We went live with our data in the cloud. We are actively going to keep pushing more and more of that data into the cloud as we continue to ensure that it is safe and secure.
Security Is Paramount
Kaufman: To that point, and, obviously, with Capital One as a bank, security is paramount, but we’ve been able to do the same thing—move a lot of our applications and data to the cloud and find a secure way to do it. From a transformative perspective, how is that, once that journey is further along, going to enhance the customer experience, for individuals or corporations?
Pandya: As we start to create the structure around those data, as we start to create and make sure those data are available to the business users, it is equally important for us to make sure those data are available to the individual and the corporate. What is our future state? Our future state is real-time tax processing. There is no real reason why, as individuals, as corporations, we cannot have access to the data and manipulate the data as we need to in real time. That is our target state, and that’s where we’re headed. That’s where our heavily focused technologies are talking about designing and architecting the environment so that, if we need to make a change to a 1040, if we need to make a change to your W-2, whatever it might be, where we are headed is the ability to do so through an online portal, without having to refile your returns. And that’s why the structure and the foundation of the data are so critical. As we modernize our individual master file, our business master file, as we start to take away from what we currently have, which are legacy extracts of data and focus on deploying APIs [application programming interfaces], the intent is for those APIs to then become real-time and accessible to everyone.
Kaufman: So, for example, I know you’ve made some good progress for individuals, even though they’re hard files. But being able to get downloads of their prior transcripts, sometimes to make payments, to make requests, you foresee similar activity and beyond for corporations, too?
Pandya: Absolutely. So, on the individual side, we’ve done some remarkable things. Certainly there’s a dichotomy of how I work within the IRS and as a taxpayer what I want to be able to see and access from the IRS. I know what my personal experience is when I work with or when I’m trying to access something on the financial side or on the commodity side. Several years back, I had a Capital One bank [account], Capital One credit card, Capital One money market—I had multiple logins, right? I had to go into an individual account and log in. That has now been centralized. I now have a single pane from which I can access all of my accounts. We at the IRS are doing the same thing from an online presence perspective: one single pane of glass for all of your account access, a 360 view. If you want to check the status of your refund, if you want to check the status of your child tax credit, your stimulus payment, [if] you want to submit something, that is already actively in place right now . . . your ability to request emails rather than to get paper notices, etc. On the business side, similarly, the idea is the same concept: the business online account. We love our acronyms, so we call them “OLA” and “BOLA.” The business online access is the same concept. It should be no different an experience from how all of you and all of us are accustomed to accessing resources. Go in, one pane of glass, to get all of your services at the IRS.
Kaufman: That’s obviously really exciting, particularly for those of us who in our team spend hours on the phone at the service center trying to move a payment around. So, you anticipate the real-time to include those type of things—basic tasks, access to accounts, etc.?
Pandya: Absolutely. The real-time is for individuals and for corporate. And you talk about spending the time on the phone—we are very intimately familiar with the concerns and complaints that people had about the calls and the frustrations. We have been, from a technology side, pushing and deploying technology to help alleviate a lot of that. In fact, seventy-two percent of our 1-800 numbers now have customer callback. So, now you can call in and say, “This is my place in line. I’m going to hang up and I’ll be called back.” We are going to have the platform ready by the end of December to have 100 percent capability in place. So, we are not only working on the customer callback from the individual side, but also on the business side. Similarly, we’re deploying—and we did this for the child tax credit, we did this for the economic impact payment—chatbots, voice bots, the ability to access the services in multiple mediums. Whether you call it omnichannel, multichannel, the ability to do across the board in how you prefer—whether it’s your app, whether it’s your chatbot, whether it’s your online live agent, live assistance—all of that is what we have been pursuing. A lot more of that focus has been on the individual side, but the corporate side has been identified. And, of course, we are actively working on deploying that, too.
Artificial Intelligence
Kaufman: Along those lines, particularly with
the chatbots and that type of technology, how are you looking at artificial intelligence [AI] to replace the call center experience with a real-time technology experience?
Pandya: That’s a great question. Artificial intelligence/machine learning is something that we have been touching on. I would say that we’ve barely scratched the surface. It’s one of those areas that is such a broad category that we want to make sure we’re doing it as it best benefits not only the IRS but [also] the individual or the corporate taxpayer. It’s similar to the cloud, right? When [the] cloud came on the scene, everybody was just talking about the cloud as a whole. Certainly “cloud first” is our mantra, and I’ve made sure that my team and the IT organization as a whole added on the cloud “smart.” We want to make sure that what we are deploying is what we are going to see the best return on investment for. AI/ML is similar in that sense, where it’s such a broad category where it can really deploy and touch on so many different things. We want to first identify where the pain points are and deploy AI/ML against those pain points. In fact, we just brought on, on April 26, our first chief AI architect. The chief AI architect will be responsible for not only identifying the opportunities that exist from a technology perspective but also actively working on and with our internal business partners to identify some of the challenges and requirements and opportunities from the corporate- and individual-side perspectives. Before we got there, however, underlying that is the data platform that I touched on. AI and ML have a pure dependency on data as a whole. So, the enterprise data platform is the foundation that is going to now enable the AI/ML efforts.
Kaufman: From that perspective, how do you upgrade that data environment? Because I know, again, in our environment as a bank, we spend the majority of our time trying to create a pure data environment that is real-time and easily explorable, etc., with all that. Talk a little bit about the challenge you have just because of the age of some of that data.
Pandya: As I mentioned, we have somewhere between twenty and thirty petabytes of data—data we produce, data we consume, data we certainly share with externals. External entities include many of our government agencies, brother and sister agencies. The challenge is creating the structure around the data and the underlying infrastructure necessary for us to create that structure. The foundation of it is so disparate and siloed that we are likely to see redundancy across the board in how much data we have without having the right access to the data when we need it. From a data perspective, our mantra has been “the right data for the right individual at the right time.” And when we talk about [the] individual, it’s individual internal, external, corporation, all of it, but it’s about the right data, the right individual, at the right time. The challenge we are seeing and have seen, besides the volume of data that we have, is the ability to have some approach in which we can have access to and create the underlying infrastructure, make the infrastructure available. And that’s where the cloud comes in. The elasticity of what the cloud can provide, as much compute-on-demand as we need and want, that is the challenge. And that’s why we’re leveraging more and more of the cloud. That’s why I wanted to put a plug in for our first data in the cloud that happened last week. We know that the constraints currently faced by the IRS are our abilities to get access to infrastructure as quickly as we need and to funding as quickly as we need. Procurement is certainly one of the areas that takes a while, not because of the organization itself, but simply because they have to do all of the necessary work before we can procure. The cloud solves a lot of that. We can now deploy to the cloud as quickly as we need to, cutting away from the need for us to now procure and deploy the underlying infrastructure first.
Kaufman: Obviously that’s the same journey we’re all on. You have heightened security concerns with the IRS, and it’s the most sensitive type of data. As you move more and more to the cloud and create those capabilities, how do you manage those concerns? What have you done to keep Congress and everyone else who’s worried about that data comfortable that there isn’t a risk there?
Pandya: As I’ve said, probably, and I’m sure I’ve over-said this, but security is first and foremost in everything we do, regardless of whether it’s something internal or external. For us, security will always remain in the foreground. It’s never an afterthought. There is no option to circumvent anything, regardless of the ability or desire to expedite anything we’re delivering. Security will always take first place. Just to provide, again, additional context, we see annually somewhere between 1.3 and 1.5 billion unauthorized attempts against IRS data—1.3 to 1.5 billion. That works out to about three million unauthorized attempts a day, just to put that out there. We know there are bad actors out there. We know that they want information. They want access to the data that we are responsible for. We will only move after we have fully validated that what we are moving is completely secure. That’s why our modernization plan 1.0—2.0 is coming out a little bit later this year—has a cybersecurity pillar. We are not putting that as an afterthought. We are calling that out. And it includes everything from extended logging to auditing to all of the security measures that we need to put into place for every single application that we deploy, every single application we monitor, every single application that we allow any internal or external users to access. So, we are doing anything and everything with security at first. Our internal process is to start with security first, and then go through the process of deploying an application.
Kaufman: That sounds great to me. Can you talk about your vision on the corporate side? We’ve had a really good start to our working group. It’s something kind of new and unique. How do you view that interaction going forward and where taxpayers, TEI, and in particular our members can partner with you and help move the technology journey forward?
Pandya: Absolutely. Again, I appreciate the opportunity to be here. I have sat in on multiple sessions, and before revealing where I’m from, I ask questions about what are the challenges that many of you might be facing with the IRS. Not because I’m embarrassed or ashamed to work with the IRS, but I love hearing what the challenges are before anyone knows that I work at the IRS. And I would love to continue working with many and all of you, really, to [show] what we can do from the IRS side to create a better experience for all of you. From the corporate side, we know that a lot of the experiences that we have been focused on and continue to plan for the future state is much more on the individual side. There is a lack of awareness, I’ll say, from a technology side, right? So I don’t want to throw anyone else from the IRS side under the bus. From a technology side, what are the challenges? When Brian and I started talking and some of the few different individuals had joined us, I did not know about the challenges with large-file transfer. I mean, it’s something that we were able to solve once we realized what it was. Similarly reading rooms was another area that we touched on and solved. Reading rooms is the ability to have access to documents and critical data without being able to download it, access it, manipulate it, etc. But that challenge was something that we were unfamiliar with, we being the technology organization. So my ask [is], I would love to hear what those challenges are. I just came from a conversation where there were discussions around the K-1, K-2, K-3, and the gentlemen on the stage were talking about, “Well, it’d be great if IRS could only standardize.” That was the first time I had heard about that challenge. So, I invite all of you to join our group and really discuss with us what the challenges are that we can solve. What we want to do is provide that same type of experience for the corporations that we offer the individual—one-pane stop, right? Online access to every service that you need access to—transcript access, whatever it might be—but I need your help. What are the challenges? Because for a while we had been guessing on the technology side—I’m going to keep saying it—not anyone outside of the IRS, but the technology organization. I would love to hear about what you are challenged with.
Kaufman: From our perspective, obviously at TEI, we really appreciate the invitation, and that’s a great opportunity, because I think we have similar goals from a technology perspective. Obviously, the IRS has a compliance function as well, but I think what we’ve been talking about is really technology in exchange for information and interaction. So, I would invite anyone in this group who’s interested in joining our group, please contact me or contact Pilar [Mata] or anyone else on the TEI staff, we can get you. We’ve been doing video calls every couple months. We’d love to have you involved directly. Or if you have issues that interest you, specific technology challenges that you’ve had in interacting with the IRS, you can reach out to me or to any of the TEI leadership, and we will get those messages to Kaschit and his team. They’ve been great so far, and we’re really excited about the possibilities of the relationship going forward. Anything you want to finish with, Kaschit?
Pandya: No, I just wanted to say thank you, Brian, for the conversation and thank you for the invite. It’s been an absolute pleasure being here. And, again, I would love to hear what all of you are talking about, what all of you are discussing [and] are concerned with and the challenges that you’re experiencing, and where and how I can help, at least from the IRS side. As a technologist myself, you know, it’s been thrilling to hear and talk about the APIs and data transfers and use of AI and ML. But I do invite all of you to share with me, whether it’s today or by joining the group, what we can do from the IRS technology side to help make sure that the corporate or the individual side has the same level of experience that we’ve been having.
Kaufman: Thank you so much, Kaschit.
Editor’s note: To get involved in TEI’s IRS Technology Working Group, please contact Brian Kaufman at [email protected] or Pilar Mata, TEI’s executive director, at [email protected].