On this episode, we showcase the transformative work of DXC technology, a leading global IT services company on a mission to drive digital innovation for customers worldwide. We'll explore DXC'S journey with AWS prototyping to harness the power of Amazon code, our customizations to enhance developer productivity while conforming to DX C'S high coding and quality standards. Well, I am excited to welcome Greg Todd, America's practice lead for data and artificial intelligence from DXC technology. Greg, thanks so much for being here with us today.
Happy to be here. Thanks for having me. And also joining us from DXC, we have Ramachandra Murthy, chief AI architect Murthy. Thanks so much for being here too. Thank you Sara, for inviting me. Here. On the AWS side, we have principal solutions architect, Asif Fouzi.
Asif, thanks for being here. Yeah, thanks so much for having me. You're really excited. And from our worldwide prototyping team, we have prototyping engineer Patrick O'Connor joining us again, Patrick, always great to see you. Thanks. Great to be back. So Greg, maybe for our international audience, can you share a little bit about what DXC technology does and a little bit about your mission? The way that we operate, um, globally, we have two separate organizations.
One is our global infrastructure services, the other is the Globalist Business Global Business Services. Um, we are a part of our global business services. So, uh, we're Murthy and I, I focus is in the data analytics and AI space. Um, the core of what we do is, uh, project work, uh, consultative work, um, helping our clients to adopt and employ new technologies that are relevant to their industry.
And quite a large organization. I understand globally global presence, is that right? Yeah, absolutely. So we have probably about 130,000, uh, depending on the day, uh, employees that are around the world. Uh, we operate,
uh, predominantly through our global delivery network. So there's several centers around the, uh, the globe that we operate in. And, uh, we obviously have our main markets, which we have in the United States or the Americas. So we'll do Canada, the United States, Latin America, south America, we have Europe, uh, and then we have Asia Pacific.
And Murthy as, uh, chief Architect, uh, for artificial intelligence. A lot of excitement, uh, lately. In your role, I'm sure. Um, can you tell us a little bit about what you were trying to achieve in this particular endeavor where you were looking at creating efficiencies for your development staff? As part of the DXC, we have, uh, what is called as an office of the ai. So we are looking at, um, activities that could improve our productivity within, as well as create the use cases that can help the customers. So one of the key use case we were thinking about it is as we are seeing the, uh, explosion of the tool, especially in the area of the, uh, AI assistance for coding, uh, it could be the GitHub code pilot or the the Code Whisperer or many such tools.
We decided we needed a productivity tool that can help us improve our productivity. And, uh, that's the initial point that triggered that how do we bring in a, a productivity improvement tool? What can concur to the DXC standards for coding the policies and other, uh, activities? And this is the place where I reached out to, uh, AWS. So Asif, you of course support, uh, DXC and their cloud journey. Uh, can you talk through a little bit about those conversations, uh, and, and how we were trying to help them not only improve productivity, but also maintain their security and quality bar? So Murti, uh, approached me with this case. His initial ask was to help him, you know, identify the right foundational model that can be fine tuned and trained on, uh, DXC'S code base so that it aligns with DXC'S standards and, uh, their, um, internal APIs and accelerators and all those things. As I began investigating, I found that there was a feature in preview at that time as part of the Code Whisperer service, and it was called Custom Code Whisperer.
And that feature was exactly built for this use case. Uh, it was the capability that would help customers get coding recommendations based on their internal code basis. So this was a much simpler approach. I mean, uh, for those of you who know generative ai, these models are very huge and deploying one and training it on your own data is, is pretty much a heavy lift. But this approach is a managed approach and, uh, was simple.
Murthyloved the idea and we decided to pull the trigger on this. The next step obviously was that, you know, how do we build and prove out the concept? So with that in mind, I started looking internally in AWS towards our investment programs, uh, that we have to help, uh, accelerate customer journeys. And I found the Worldwide Prototyping team and they love the use case, um, and they wanted to partner with us. So we all came together and we decided to build out this, uh, pilot. Mm. And Patrick, of course, exciting for you as a prototyping engineer to get to work with services and features that are in preview. Um,
talk us through a little bit about how you approached the engagement, the team. What's exciting about this technology or or when it came to the Prototyping team is the Code Whisper, not just the custom part, but the Standard Code Whisper. This was the first of its of its kind a coding assistant prototype that we were excited to, to jump on, explore some of the productivity gains and experiment with the coding assistant, which has been taking off Prevalently for the last year. Uh, so the approach we had with this prototype, given that it was the experimentation of a feature set of Code Whisper, was that we designed an end-to-end experience that allowed D XE to ingest data, curate the data through several scans, output the data, uh, and then eventually get to that customization where we would actually work hand in hand with field engineers rather than us showcasing what Code Whisper can do is we put it into the hand of DXC engineers and we have them technically validate.
Right. So you approach the engagement in two separate sprints, is that right? That's correct. The, the prototype we were facing with here had two special aspects to it, and we've spoken about it before, is the first section is focusing on the code assistant element, which was the core focus of Sprint one. And then the second element was fine tuning it to DXC information.
So part of Sprint one, um, proving out Standard Code Whisper is we actually sat with some field engineers, the the typical, Hey, here's a coding example, go solve it with CodeWhisper without it, we were able to get some really exciting results, not just from our side, but the engineers themselves were getting quite excited with how they were able to speed up. So the first sprint tick, we were able to establish the code assistant element, and then the second phase, which is Sprint two, we then dialed it up to not just be standard, but to be specific to DXC pattern or DXC code base. And going back to that first sprint, Greg helping cultivate that adoption and fostering the adoption, the the sort of learning by doing so important when, whenever there's a technology disruption right, to upskill folks, but bring them along, um, and have them experience it in the journey for themselves. Yes. Yeah, absolutely. So what's big with us, um, is the intelligent augmentation of our workforce and our client processes.
So we're adopting generative AI not only for our clients, but also for ourselves. So that level of productivity that we can get through the use of these new tools is super important for us. That's the future of IT services. So when we thought about how we'd wanna do this, we thought, who can we identify as a partner who shares the same vision? A AWS was on that short list and the customization of Code Whisperer, that use case that we were able to produce around having our coding policies and our standards intelligently augmented, uh, for our developers was super important for us.
Right? And Marti, that investment that you've made in your Heritage code base, your APIs, your libraries, uh, that you already have, so important for maintaining the consistency and uh, and quality of the code that your, uh, engineers and, and architects are developing. So if you look at it right now, over the period of time we have been working on developing the code and each time we implement what kind of improved standards we have to have it, so you don't wanna start all over again. We wanted to use those standards that we have built in into the next phase using this coding assistant. That's why it is all the more important for us to use the coding assistant with the coding standards that we have built in. Right. And Patrick, take us through a little bit of the architecture for how you set up that second phase when you're gonna actually go and, and customize, uh, for DXC specific code base.
The architecture itself can be broken up into two segments, the customization element, but just as important is how the data makes its way to the customization to then train. So starting off, the journey of this architecture is we had DXC engineers collect, um, code from their code bases and provide it into an AWS account. Now, given the time constraints of this prototype code whispers supports a series of languages, some include Java, TypeScript, JavaScript, and Python.
For the purposes of this prototype, given that it is time box, we focused on the one language of Java. So DXC was able to provide 300 megabytes of DXC specific code into the account. We then created a pipeline using Code Build to then run a series of scans on this code.
The reason for this was to provide quality assessment reports, uh, generated by Soner Cube, invite the human to not just trust the code and put it into customization, but to validate it before it comes in. That's the first segment. The second segment is then connecting code Whisper customizations. This is done through a series of services. Some of those services include the Code Whisper service itself, but also for identity management. We use AWS Im identity management to then connect external users via their IDE into the customization. So we train the customization on the cloud, but then from the DXC field engineer's perspective, they log in from their VS code and they have access to these customizations in real time.
Mm-Hmm. Visual Studio. Yep. mm-Hmm. Correct. They're the IDE that they're using in their day-to-Day work. Right. This is just one of many ides, um, that you can use. And then that customization. What were some of the learnings that you had out of, uh, working with the engineers and Yeah, how did you go about doing that validation step? We actually had quite an swarm and a diverse range of engineers.
Some included sa uh, solution architects that were in the field providing advisory services. But then on the other side, we actually had some field engineers who were building the code base that we used in training at. They were also dispersed across the globe. So we had the majority of our developers within the Americas, but we actually had some field engineers within the Asian region as well. Some of the feedback we collected from these engineers ranged in cutting down time of tasks that would normally take half an hour of development. Bear in mind that a lot of these time blockages come from Googling, you know, basic activities such as how do I do this? Google It, figure it out, bring it in, test it in. But with Code Whisper,
what you can actually do is you can just simply write a question, have a provided code base, and then work backwards. So some of the feedback we collected from the various team members included a cutdown in time expertise in code bases that they might not necessarily be familiar with. And at times they were quite excited for their output With this tool, they could produce newer things that they may have not been possibly able to do before. Right. And Assif, of course, this is the first time you're seeing a code whisper customizations in in action, some of those key features that that help with productivity. Yeah.
Yeah, definitely. It was very exciting and some of the feedback that we were getting from the testers and engineers, they were very encouraging. I mean, I think there were people who were involved in the testing who were not really familiar that much with Java, and they were, uh, writing great Java code because of this tool, for example. Right. Uh,
that was very encouraging, but also from my perspective and from AWS's perspective, working with a partner like DXC validating the, uh, usefulness of a new feature for our customers, it's a very important step. We are very customer focused and we work backwards from customers. So this was also helpful for DXC as a customer. And, uh, it's, it's a very important, uh, feature that we are launching. Absolutely. And, and Greg that upskilling mentioned the, that, uh, some developers who might not be familiar with a particular language, creating those opportunities for people to learn and to grow in their development, guessing it's sort of fundamental to, uh, to your business needs.
Absolutely. So we are a people business. Um, the IT services that we provide, we wanna make sure that our individuals who are working in our delivery organizations have the ability to grow their skill sets, adapt to new technologies, and provide new solutions for our clients. So I think this project was a great example of us providing an AI augmented service, uh, with AWS it enabled speed and efficiency on how we do our coding, and I think we have numerous opportunities to improve automation and innovation, uh, with partners like AWS. Fantastic. And you know, Murthy, a lot of times in this podcast we talk about experimentation and failure or blockers being sort of inseparable twins, right? You can't do something without there being something new, without there usually being a challenge that's in your way. Uh, was there something particularly here that was a challenge that you had to overcome or a blocker in the way? There was one particular thing we were, um, always concerned about, uh, as we are partnering with AWS and, uh, one of the key requirement was I, we needed to provide, uh, 300 MB of the code base.
And as you know, it's an IP that we own it. And, uh, suddenly the question starts raising in everybody's mind, is this code going to be in the public? Will uh, somebody be able to see all of our ip? Uh, how do we make sure that none of these thing goes out? So there was lot of concern around this one, which made us go back and forth with the actual business owners and the AWS to prove a point that this is all within the secured environment and, and, uh, we manage that one. So Asif, good news from the Code Whisperer customizations on that front, right? Oh yes. Oh absolutely. And you know, as everybody knows at AWS security is Job Zero and we have certain guiding principles, you know, use least privilege and nobody at AWS gets to see customer data, right? That's how we operate and that's all baked into our DNA. So no different here with this. In this particular case,
what we did is that we decided that we'll do the actual pilot building in a DXC owned account. So now the data is contained within an account that is owned by DXC. So it's not leaving the account. The data is never going to be used by the Code Whisperer service itself for any purpose. Its only job is to be able to provide, uh, recommendations that are aligned with DXCs coding standards.
So we did that. We used a DXC owned account. The second thing we did is that we followed the principle of least privilege in giving, uh, our prototyping engineers like Patrick restricted access to operate the account, but only to be able to do what they need to do. There's no admin access here.
Very concise defined roles on what kind of access they have. In that. And that's very typical, isn't it, Patrick? For, uh, prototyping engagements in general, ensuring that we have that secure environment that we're working in. Spot on. We have various different models here in prototyping. Sometimes we build on our side, sometimes we build on their side. We try to work backwards from the customer in designing our model or designing the engagement to best accommodate as if Medi spot on in the sense that data is a core element on AWS, it's contained.
Code Whisper is no different to any other service in that we respect, we secure the data as best we can. And as if hit the nail on the head in that when we designed this prototype, we as the prototyping engineer had restrictive access. So we were granted only the privilege to jump into their account. The data never left their AWS Sandbox account and anything we were doing within their account was being monitored, praised, and with minimal privilege. Mar, where are we now? What's next? After you have a group of folks who have tried this out, you've tested it in the field and verified the efficacy of, uh, using Code Whisper, what's next, uh, in terms of your journey. The outcomes of this? It's a, a successful results.
We were able to show it in the office of the ai. So now the message is clear to many of the business units that this is a possibility. We have a track on how to do the customization on top of the Code Whisperer. We did it with Java,
but we have different languages and different business groups that they use it for internal work as well as for the client's work. So we will be working with the different groups to build the customization for their particular need. Uh, sometimes we can do it our own based on the guidance that is given by the A Ws team because they have given, as part of the prototype, uh, we receive the deliverables as the entire flow on how to do it and all. So we can do it ourselves, but if we are stuck, we're always, uh, reaching out to the AWS to provide that, uh, support for us. So we are looking to implement this customization within the different groups to improve our productivity so we can be competitive in the market. Well, so exciting. Greg Murthy, Asif Patrick,
thanks so much for sharing this journey, uh, with us. And uh, as always, we'd like to leave off with some reflections what you learned or what you would share, uh, from, from those learnings in this experience. I'll, I'll start with you Patrick. My reflection is really on the past year of development, particularly as a prototyping engineer is we've seen an explosion of Gen AI applications and I really think, in my opinion, that we're going to see further evolution of Gen AI application and use cases going into the future. What this prototype provided is just how effective and impactful one use case of gene AI is. So in this case,
we used Code Whisperer to operate as code, and we've heard from this prototype of the impact it's had. I think going into the future, we're gonna see a lot more use cases, uh, spur up a lot more prototypes, uh, which just shows me, you know, from a reflection point, get on AWS start to experiment with these gene AI technologies. Um, and then at least you're ahead of the curve or on the curve of innovation, uh, particularly as new features and services evolve around this exciting space. Fantastic. Asif, how about you? What were your learnings, your takeaways. The power of the technologies? Undeniable. I mean,
I believe that it's going to be embedded in every industry across all business functions. And sitting through this project working with, uh, DXC and the prototyping team, it was really amazing to see the ROI right within two sprints, we were able to generate that wow factor with the developers and engineers who were testing it. And that is different from what we have seen with previous Breakthrough Technologies. This technology has immediate impact and I would encourage people to learn more about it, experiment, whether it figure out ways of how to embody it in their day-to-day work.
Dorothy, how about you? What were some of the reflections or takeaways that you had from this experience. This team? Fantastic. Uh, we've been working, uh, with AWS for a long time. And, um, as we did this prototype team like Patrick and Asif helped us navigate, um, every step of the way to make this one into a successful project. As I observed the, the, the gender A TVA explosion in the last year, uh, we we heard many times in, in places like people with an AI or the employees with an AI has an advantage as against with the employees without now are seeing companies with an AI has a huge advantage as against the companies without AWS is helping us in making that company with an ai. We are building many new use cases and this has given us a huge advantage to build even more and especially with an AWS team.
Thank you. Absolutely. And Craig, final words from you. The focus that we've had on intelligent augmentation of our workforce is, uh, important paramount for us to be competitive in this marketplace going forward. I think this is a great example of AWS and DXC working together on the future of digital. Um, so excited to continue working with them and bringing even more use cases to, to light. Fantastic. Looking forward to, exciting future ahead of all of us.
Uh, so thank you all for being with us today.
2024-06-03