Impactful work: Helping developers around the world improve productivity with AI

AWS announced Amazon CodeWhisperer at re:MARS—Amazon's global AI event for machine learning, automation, robotics, and space—in 2022, one of four innovations from AWS that will make generative artificial intelligence (AI) more accessible to anyone who wants to use it. Behind this innovation from AWS—which offers the broadest and deepest global portfolio of AI and machine learning (ML) services at all three layers of the stack—was a large team of AWS builders and partners imagining new ways to make customers’ lives easier with generative AI.

CodeWhisperer is an AI coding companion that “radically improves developer productivity by generating code suggestions in real-time based on developers’ comments in natural language and prior code in their Integrated Development Environment (IDE),” as described by Swami Sivasubramanian, AWS vice president of Database, Analytics, and Machine Learning, in a blog post announcing AWS’s new generative AI tools. In an early productivity study, CodeWhisperer helped developers complete tasks 57% faster, on average, and those who used the tool were 27% more likely to complete the tasks successfully than those who didn’t.

“This is a giant leap forward in developer productivity, and we believe this is only the beginning,” according to Sivasubramanian.

As AWS innovations such as CodeWhisperer open the door to a new world of productivity and creativity—it’s free for individual developers globally—they also inspire former, current, and future engineers to reimagine what it means to think big and make a difference in the world.

We chatted with some of the builders behind CodeWhisperer below—Ankur Desai, Srini Iragavarapu, and Parry Bhatia—who share how innovative thinking, trial and error, and persistence helped them create a tool to benefit developers’ lives, as well as their own professional journeys.  

Ankur: People. Purpose. Persistence. People—being developers in this instance—come first. The journey was about solving developers’ challenges, and as developers ourselves, we have a deep empathy for these challenges and passion for solving them. We also needed to focus on a singular purpose, which was improving developer productivity, ensuring they are more satisfied in their jobs. Finally, we needed persistence to get through the setbacks we would face during the journey.

Parry: Training these models and running inferences at a massive scale present unique challenges. We often found ourselves challenging the traditional approach to developing new services, and that’s a place where a lot of interesting ideas and innovations came from.

Srini: Developers are tough customers. It takes a lot to impress them when it comes to cutting-edge technology. And really, there are few things happening in the world right now as cutting edge as generative AI; and that is space travel and autonomous vehicles. And as I look back on the journey, I don’t think I’ve ever worked harder than I have in the last two years. We have an absolute killer team—one of the best I’ve been on—all putting in top-quality hard work.  

Srini: In this day and age when the software industry is radically changing and people are switching jobs often, there really isn’t much connection. Yet in the last two years, across our team, we’ve celebrated new borns, weddings, and promotions—it’s this kind of connection on our team that’s rewarding. And when Ankur and I are going at it, disagreeing; or Parry and I are disagreeing; or leadership comes in requesting certain things, it’s in these moments—all the way from leadership to individual contributor—where we knew we were all in this together, operating towards the same goal.

For me personally, a middle school friend messaged me on LinkedIn when we announced CodeWhisperer in preview. He was like, ‘Whoa, I didn’t know you were a part of this team; that is so cool; is this going to support C++?’ So, when somebody like a middle school friend who I have not spoken to in 25 years randomly reaches out and let’s you know that it’s cool you are a part of this service, you know you’ve done something good.

Parry: In the last two years, working on a project of this magnitude, I realized I could do a lot of things that I didn’t think I could do prior to this. I’ve gained a lot of confidence from this experience and feel like now that I’ve done this, I can do almost anything.   

Ankur: The most rewarding part is building the team and the connections. It’s that feeling of being a part of something bigger. The other part is knowing that we are working on something that will change lives for better. And this is not incremental change, this is step change for a lot of people. It’s not often that you get to work on products and solutions like this that are truly going to change many people’s lives.

Ankur: Again, I think for me it’s having that passion and empathy for solving developers’ challenges—that keeps me going.

Parry: This launch required a lot of hard work and cross-AWS collaboration, and without this great collaboration, I don’t think we’d be where we are now. Everyone was motivated by the same goal regardless of how much ownership they had. And it’s this collaboration kept everyone at pace and moving forward.

Srini: To Parry’s point, there were a lot of teams involved in this service. And the leadership team was very persistent in pushing all of our teams to keep moving forward.  

Srini: Before this I was on the AWS SageMaker Elastic Inference team when someone told me about CodeWhisperer. So, I talked to the team, and at the time it sounded cool, so I made the move to CodeWhisperer. It didn’t hit me until six to eight months in what it truly meant to work on this. I realized it wasn’t just a service, but that we’re talking about generative AI here and creating something that is going to create step change for a lot of people.

Parry: When you’re working on something for the first time there are a lot of unknowns. You hit a lot of walls through trial and error and move on. But like Srini, I think we could’ve done a bit more on the ‘think big’ front earlier had we known where we would be today with generative AI.

Ankur: I think you can see the theme here is ‘think big.’ To a lot of people, generative AI sounded like science fiction at the time. It was hard to visualize that you can input something here, and magically things will appear. If we all knew where generative AI would be today and what would be possible, I definitely think we would’ve planned for larger things. This technology just has grown leaps and bounds in the past two years and it was really hard to imagine that back then.

Ankur: For developers, I think the benefit is they get some work-life-balance back. They can focus on exciting problem-solving rather than repetitive, boring coding tasks. For customers, they get a productivity boost where they can launch products faster and can implement more ideas and features as requests come in from their customers.

Parry: I think for the cloud computing community, CodeWhisperer is going to change how developers get to their end goal.

Srini: If you take CodeWhisperer and combine it with what we are doing with generative AI and these large language models, we’ve shown that we can actually take these services, bring them to production, and make them available to engineers everywhere. We are at production quality, and the business impact is that developers everywhere can get more creative with our services, build on top of them, and create even greater use cases through their innovations.

Stay in touch  »
Tell us a little about yourself and we’ll share more of what makes AWS the best place to build and learn.