UST is a digital transformation services company with more than 35,000 employees around the globe. The company’s services run the gamut of IT, from cybersecurity and data analytics to supply-chain management and next-gen cloud infractructure, and its clients run across seven vertical industries.
Like every other company, UST has had to deal with a dramatically different workplace compared to what existed prior to the onset of COVID-19. It has moved from fully remote at the beginning of pandemic to requiring IT employees to be in the office two to three days a week.
The company is also in the midst of creating a more flexible, blended workforce to meet its own and customers’ needs more effectively. UST’s Open Talent strategy leverages the freelancer talent ecosystem. Its IT team has also had to adjust to the fact that technology automation is here to stay and will only continue to grow.
Part of that automation includes generative AI technologies such as ChatGPT, which can perform myriad business tasks, from uncovering software defects and data mapping to automating supplier invoices.
UST’s clients have also become more conscious of costs at a time when global economic security is in question and economists continue to warn of a possible impending recession.
USTUST CTO Niranjan Ramsunder
UST CTO Niranjan Ramsunder joined the California-based company in 2006 as a technical solutions manager. Since that time, he’s also held roles as UST’s director of technical solutions and global head of solutions before being named CTO in 2019.
Since becoming the CTO, Ramsunder has been focused on data, AI, and how to identify client problems — what he calls “the left side” of the service role, or discovering for the customer why they’re in the fix they’re in.
Ramsunder spoke to Computerworld about the technology industry and the future of work, IT project costs and challenges, IT culture and trends, how generative AI will change the way technology problems get solved, and how the IT workforce can prepare for the future.
The following are excerpts from that interview, lightly edited for clarity.
How has your role as CTO changed over time? “My role has been to push that envelope to the left — getting to the problem. Why do we have that problem in the first place? For example, we had one customer who was using AI to do reporting for analysis. We realized very quickly they couldn’t state what wanted to do in terms of the number of critical data they were analyzing. Then we ended up very quickly with a new build on AWS Cloud. Sometimes problems and solutions have to go together in terms of what can be solved. What technology choices are they making? Is it cost effective?”
What are the biggest IT trends? “Obviously, everybody is looking at ChatGPT. Generative AI is showing up everywhere. Normally, I’m hesitant to jump on the buzzwords because they really don’t translate to business very quickly.
“Seven or eight years back, we had a buzzword for moving to the cloud. One of the first jobs we had with that was with a company called G4S, which delivers cash to ATMs with trucks. The trucks needed to get updates on where to go next for the day. They had a device that got updated at the center where they collect the cash and then distribute it. They wanted to use Azure. That project really suffered because Azure was not mature yet. It was 2010 or 2011. So, the cloud was more of a buzzword. Now cloud is taken for granted.
“There’s a lot of discussion about… blockchains. But none of them translated into business. Right now, generative AI and large language models [LLMs] is definitely a place where we find the buzz and the reality are matching. So, that’s the number one trend in my mind for customers.
“The number two trend is a constant focus on costs. Particularly because of the expected recession from an economy point of view. From a client point of view, they’re all very conscience about spending cash. They’ll only spend cash if they’re able to see results in three to nine months — no longer than that. Every dollar they spend, they want to see it result in something valuable. Our client jobs are linked to that: spend and results.
“Those are the two trends that inform everything we do. How do we talk to the customers? How do we get funding for those projects?”
How do you see AI affecting the workplace of the future? “The reality is, for companies like us — people who work with customers to find solutions for them — we’re finding AI has become a kind of necessary evil. Without it, we can’t survive. With it, there are too many risks.
“For example, AI may give you a wrong conclusion; it may be biased in its conclusions; it may give you things you’re not able to track and audit backwards very clearly. At the same time, not using it puts you in a hole, because you’re not that productive compared to someone else using it.
“So, the use of AI is getting more targeted toward [specific use] areas where there is less confusion about hallucinations and wrong decisions, but more about, ‘Can it improve productivity with less risk?'”
How do you see AI affecting jobs? Will it eliminate more than it creates, and what jobs do you see going away or being created? “That’s a question that bothers all of us. It’s a very subjective answer. From my point of view, I believe we at UST are focused on providing solutions for customers, and so AI will create more jobs there.
“[On the other hand], the overall number jobs will I think reduce by, say, 20% or 10%. At the same time, the kinds of jobs coming to the marketplace [becase of AI] will not be the same. And not all jobs are going to go through this crush. Knowledge jobs — white-collar positions — that kind of work is going [to be reduced].
“One fundamental change, though, is the skill level that we have today will change. The kinds of jobs going away and the kinds of jobs coming in are definitely not the same. You may not [need to] be an expert in AI to use AI today, but when you get into the next level of the job change, you need to understand the implications of it. You have to understand, ‘What can I do with AI that I cannot do today without it?’
“So, for example, if you’re someone [managing] invoices and making sure payments are being made correctly to a supplier, you’re going to have to change the way you look at your work. You cannot be [manually] looking at matching invoices and paying suppliers. You can look at AI and say, ‘Can I get some ideas on which [invoice] to look for first?’ That thinking then will drive the way they act and what their jobs look like.”
How important a skill is generative AI prompt engineering to yours and other orgs? “I think prompt engineering is the key to using AI well. The way that it’s designed and the way it’s tracked is where the success or failure of whole area is going to be. That’s absolutely crucial.
“If you look at the kinds of jobs I want to prepare my child for today, I would say go for prompt engineering. The model’s deployed on hardware. What are the most efficient ways of doing it? Those are the two areas you’re going to find the most opportunity to save money and make money for an organization.
“So, when looking at jobs for [which high school students should prepare], I’d say look for understanding the implications of cost while working with AI. And look for the costs of prompt engineering, hardware engineering. They all link to the same problem: how can I use AI in a way that’s cost effective, gives me quick results, but doesn’t take me in the wrong direction?
“That’s where all of them are tending to go toward. Effective use of AI.”
I understand tailoring some of these LLMs for specific business use can be very expensive — even cost millions of dollars. Have you tailored an LLM for use at UST? What was your experience? “Yeah, we’ve been doing a lot of work with AI over the last seven or eight months, ever since OpenAI released ChatGPT and it became much more visible.
“To answer your question, yes, we’re using it for problems like if I want to generate a test score to test my software. I can do that automatically. I can look for security violations in my code. I can do it with ChatGPT frameworks — large language models. Instead of going through every line of code to find defects, I can run it through this program. Again, the risk of not getting it right is not as much as the risk of not doing it.
“Also, more mapping activities — for example, when you have data mapping activities between your source system and target for analysis. That process [normally] takes a lot of people. When data flows in large volumes, you get too many errors if it’s not done well. That process can be automated with AI.
“So, a lot of this work is where we’re spending a lot of our time in terms of using AI for productive work and reducing errors in other work. That’s really where the opportunity is. Can I be more accurate? Can I be faster? And at the same time, can it be cheaper than doing it any other way?”
Who’s going to take on the task of prompt engineering — IT workers or business-side workers? “The easy answer is both. But you need to appreciate what prompt engineering does. What kind of prompts make sense? You need to have domain knowledge but also need to have an IT framework in your mindset to understand how AI works.
“Managing AI better in terms of your prompt engineering, managing the way an AI model is built, has got to be with an IT sensitive, knowledgeable person looking at the domain capability. You can hardly do any work today without IT. You cannot understand how automation works without IT. So I would say get a domain person, train them on AI. But you need both, and they need to be trained specifically on this.”
How have you adapted to hybrid work within IT? “That’s been one of the big challenges, because with COVID people expected to work from wherever they were. Then after two years, suddenly COVID went away somewhat. Now, we want the workers to come back to the office, but the genie is out of bottle. We’re finding it very difficult to push it back into the bottle.
“And people are either leaving or they’re dissatisfied or they’re not productive. There are many things that are not unique to the way different societies operate. At the same time, a lot of people hate the commute… and spending time that’s not productive.
“The other side to that is why we want people to come back. We cannot collaborate. In spite of all the collaboration technologies we have, if I was meeting you face to face, I’m sure our interaction would be 10 times better because you can see things. You can look at body language, and how someone is thinking even when they’re not speaking.
“So, we’re finding there’s a mix between some people who do well in a hybrid [or remote] environment… and then some jobs where there’s more collaborative work needed — working with customers, working with larger teams. We find it makes sense to have face-to-face interactions there.
“What we’re trying to mandate is at least two or three days a week in the office when people work together and then have time on your own to do stuff. Because meetings are not the only thing you do at work — a lot of what you do is thinking, cohering things into something meaningful. Those things can be done in your own comfort zone; if it’s the office, it’s the office. If it’s home, it’s home. So we’re looking at a truly hybrid environment with a mix of flexibility and some days in the office.”
How has IT culture changed over the past three years? “What has become more and more clear to me is automation is here to stay — even in our work. For example, a lot of the work we were doing before was manual. Manually configuring things, setting up things, mapping things to each other, creating an environment. A lot of that is getting automated.
“So, we’re as much in the same revolution as our customers are in. That is disrupting the way we imagine our jobs, imagine the people we work with, the way we interact. Everything is changing fundamentally. Our children will definitely live in a different world than we live in.”
What IT skills do you believe are most important for the future of work? “I believe for people who are coming up today, for any area of enterprise work you need to learn how it will be impacted by things happening today in technology.
“You need to have a constant outward focus on what’s going to impact you. You don’t want to get surprised. Needing to learn [new skills] used to be a requirement for the ambitious, but now even to survive you’ll have to have that future-proofing mindset for your skillset. That future-proofing is mostly technology based. It can be based on automation; it can be based on AI; it can be based on a completely new way of doing work that comes in and can replace you.
“I don’t think anyone taking a job can be complacent in today’s world. Every job, including that of a doctor, for example. If they’re expecting to be productive throughout their career they’re finding their jobs may be at risk, because even a diagnostician working on AI may be able to offer a better diagnosis for a patient. An automated system monitoring all your heart, your bloodwork, your various vital signs can give you a better diagnosis. So you’re competing against that all the time.
“So, your willingness to be aware of what is going to take your job, that’s a new skill you need to survive in this world.”
What are the greatest IT challenges today? “I think the number one challenge today… is you’re no longer just finding a solution, but you’re having to find how cost effective it is to do it that way. How expensive is it to do it that way? It has to give you a result that’s cheaper than it cost to bring automation into it — or AI or ChatGPT or the cloud. All of them cost money. The question is, who makes the money?
“So, for example, you have cloud providers making money. Then you have the GPU manufacturers making money. Can you decide what is the most cost-effective way for my company to survive?
“So first, be open to learning all the time. Second, be cost effective in all you do. That’s going to be driving a lot of job positions tomorrow. We’re not going to be having big budgeting exercises. Money is always in short supply.”
