For nearly five months, Pavan Pidugu has been calling the technology shots at the Transportation Department, blending a familiarity with the agency’s mission as a directorate chief technology officer with a prior track record of tech success at retail behemoths Target and later Walmart.
Among the first chief information officers appointed in the new Trump administration, Pidugu’s familiarity with the department mission — and its challenges and flaws — allowed him to start “with a goal list and a playbook on day one” that’s driving tech transformation at the department overseeing and securing the nation’s transportation systems.
Pidugu discussed his approach and early efforts at DOT in an interview at GovExec’s Government Efficiency Summit in late June. Below is that interview, edited for length and clarity.
Nextgov/FCW: We’re talking a lot about efficiency in government right now. What are some of your priorities?
Pavan Pidugu: Everybody always thought the CIO’s job is [to be] responsible for network security and Outlook, and then maybe desktop support, and everybody else…they get to do their own things. The CIO is not at the table or was not even informed or aware of certain things that were done in the departments or in general. I want the OCIO within Transportation to be a technology shop where we build technology. That’s the culture we want to create in-house. We’re building applications. We’re building products, along with being everything else.
So we got the [Transportation] Secretary Mike Duffy to sign a memo directing all of the administrators and all of the assistant- and deputy assistant secretaries informing them of no more shadow IT.
We’ll have one DOT IT.
Nextgov/FCW: How many angry emails did you get when that happened?
Pidugu: I made good friends. But that’s the way to do that if you want to be a central technology arm that actually does technology right in the process. We’re also uncovering a lot of redundant systems. As an example, we have 10 to 14 different grant systems. We don’t even have 14 sub-agencies. I don’t know how we got 14 grant systems in the first place, right? So unifying all of the grant systems into one common product across DOT is going to help us be able to have that visibility.
Also, if I am a roadside truck inspector, there are 14 different systems that I have to go through to finish one truck inspection. That seems inefficient, it is painful. I really felt [this] a few years ago when I went to visit one of our roadside truck inspection sites in Dumfries, Virginia. The lady officer was sitting in a crammed car trying to type, toggling back and forth between different systems, [and] it was 105 degrees out! It’s painful and something to be ashamed about as a CIO or CTO back then.
So we’re trying to simplify that and give a one-stop shop product. I’m supposed to do my job, all the capabilities that are required to do my job as a federal employee are all in one place. We have this in our day-to-day lives today, right? Imagine a world where you have to go to five different Netflix apps to watch a movie. I would never watch a movie if that’s the case, right?
Nextgov/FCW: It sounds like you’re trying to make DOT into a product organization. How are you driving that change? There’s more to it than getting the secretary to send a memo.
Pidugu: Yes. You simplified it. First and foremost is helping people realize it’s a problem. Everybody wants change as long as it doesn’t involve them. So trying to ensure and provide that comfort to people that they are the first ones to lead that change, and everybody marches behind them, it has been the trick.
The other big thing for me was trying to learn who influences who within the department and trying to sell that change to the people that can influence the rest of the organization. Because, like you said, memos are great, but the work has to come from the heart, like the change, and that motivation has to come from within. So for anything of this magnitude to be successful, mindset is key.
Second is actually having to understand what capabilities are required, where data is, where data flows, who changes data, all of that. And sadly, some of these applications, or some of the technology and processes were built way back when. Nobody knows why they are that way. So sometimes it is okay to break things to know what the underlying plumbing is.
So we’ve taken some broad choices to say, ‘Okay, I’m going to shut this off for a day and see who screams the most,’ and have those conversations.
And then last but not least — which is the easiest part — is actually building tech. Getting an understanding of what the change is and how you’re going to drive the change and two, making sure people are upskilled.
Processes, people and products — the three Ps in my dictionary.
Nextgov/FCW: Do you have an example of change that happened when you ‘turned things off?’
Pidugu: I’m going to harp on SafeSpect, which is probably what got DOT to understand we can do technology products in less than six months. We did a lot of experimentation with our [commercial motor vehicle] inspectors and it started off as a pilot where we just took one type of inspection that is done on a truck, and went live in 120 days. Acquisitions [previously] took seven, eight months. That was a good example. I used to tease my acquisition leadership partners to say, ‘hey, we’re building in three months, you guys need to speed it up.’
That gave good motivation for the rest of the organization to realize we can do it. But in that pilot, what we were struggling to do was [understand what actually happens after the inspection data is in our Motor Carrier Management Information System]. It goes 18 different ways, and we only knew seven. This process helped us to identify those 18 ways, and we built the additional APIs that were required to do it. But that’s the reason we picked one inspection type versus all seven on day one.
Nextgov/FCW: That ultimately speeds up these inspections, which benefits your employees, makes their jobs easier, and frees up other time to do things?
Pidugu: And not just our employees. Also think of the wheels not moving. It’s a cost for these companies. It’s also 30 minutes of time that the truck is sitting is 30 minutes of time somebody is waiting for their groceries or medicines or something.
A funny story in that, as we were rolling that out, some roadside inspectors had an internal competition amongst themselves [comparing the new system to the old way]. They go under the truck, check the bolts, nuts, everything. It’s a 70% time reduction. It helps the drivers go farther, companies reach destinations quicker, and customers receive their items quicker.
Nextgov/FCW: Give me an example of something you are building, have built or want to build that could improve operations across DOT?
Pidugu: I’ll talk a bit about AI, because everyone is interested in it. We were having conversations with some of the new appointees and there were notices of proposed rulemaking [NPRM]. And people wanted to know, is there a quicker, better way for us to segment these comments in some groupings? Some NPRMs get hundreds of comments, some get thousands or tens of thousands, so it’s a monumental effort to read through every single thing.
So we built an AI analyzer tool in two days, and then we showed it to some of our regulatory people in policy and rule-making organizations. They gave us [four pages of feedback we incorporated in one afternoon.] We met them again in the morning and by tea break, we were able to have it deployed in our dev environment, and we got more feedback.
So, an AI analyzer that can go and review 6,000, 7,000 comments on an NPRM or a tweet and be able to provide a summary, synthesize and categorize all this — if we can do that in two days, in four years, we can do a lot.
Nextgov/FCW: What’s the right approach to AI? Build or buy?
Pidugu: I think if you listen to Secretary Duffy or even the president, they’re all about building; they like building things. We are in transportation. We build highways, we build trains, we build train tracks, we build transportation safety, and we build bridges. We are building ships now with [MARAD], so we’re building all the things. Why not the technology that we use?
So we will build, we will create our own large language models for those things that we need to have full blown authority on what the AI is generating and why. And there are certain things that are common enough, commodity enough, where I don’t need to know what the LLM is behind, how it is built, or what the guidance is for that model.
Nextgov/FCW: What’s your advice to the GovCon industry?
Pidugu: I would like to give one message, but don’t be offended. There needs to be a lot of change in thinking in the players here trying to solve government problems in lighting speed.
I think when I first got here, all I saw was like, ‘Okay, how many years of base-plus is this contract going to be?’ And we have time to build it in that base-plus, right? I don’t want you to come to me with that mentality, and I’ll be open, but if that’s the mentality, that will probably be the last conversation we’ll have.
Understand our problems. All of our problems are not hidden secrets, they’re all public. Whatever we’re trying to do, we’ll have RFQs, RFIs and everything out there. Understand those problems and come in with solutions on how to address those at lightning speed. I’m all about if I can’t do it in three to six months, I won’t do it.
Nextgov/FCW : Before we conclude, I want to address the Department of Government Efficiency’s impact as well as the new energy around the Federal CIO Council.
Pidugu: Speaking for DOT, the representatives that came from DOGE have been employees of DOT and they have been partnering with us to identify opportunities for us to save money and get better. I won’t name the product, but we had the same product procured within DOT five different times…So that collaboration with DOGE has been helpful to identify some of these things. We have gelled well.
For the IT employees and staff who I meet regularly with, there were concerns in the beginning and as much as we were able to communicate, share and what they saw — I think it’s important to have the words, actions and results be aligned. When they saw the words and actions in alignment, the trust was there a little bit. Trust is not [a] one time thing, right? It’s an every day thing. So, there will be more.
And the [Federal CIO Council] has been great. I’m so glad there are leaders like [Federal Chief Information Officer] Greg Barbaccia and a few others in that council that are hungry for change, and they have done similar things in their past lives, right? CIOs can be two ways: A great policy person… or you’re going to do that, along with it, you’re going to be the technology product delivery organization, which I love to be, and what we’re aspiring to be.
Some of these processes that were built many, many years ago need to be changed to help CIOs like me to go faster. So the council has been great. It’s closed-room, in-person meetings and [a] lot of good discussions happen there. A lot of actions are being taken out of those discussions.”
