Artificial intelligence (AI) is advancing at a staggering pace—but it wouldn’t be possible without open source. In a compelling interview with Swapnil Bhartiya on TFiR, David Nalley, Director of Open Source Strategy and Marketing at Amazon Web Services (AWS), breaks down how foundational open source tools and philosophies are shaping the future of AI.

From PyTorch to LangChain, open source software is embedded at every level of AI development. “People are using PyTorch to do the training. They’re using things like LangChain and Crew AI to string models and agents together,” Nalley explains. These aren’t niche tools—they’re becoming the core building blocks of generative AI (GenAI) and LLM infrastructure.

But as the open source ecosystem powers AI’s growth, a major question looms: what exactly qualifies as “open source AI”? Nalley cautions that while many models are now released with more permissive licenses, most do not include access to the original training data. “Without the data, it’s really hard to be able to recreate and modify,” he notes—an omission that violates foundational open source principles such as those laid out by the Free Software Foundation and Debian guidelines.

This missing piece creates confusion for developers and organizations alike. “There’s still a little bit of controversy,” Nalley says, as different groups—including the Open Source Initiative and Linux Foundation—work to define what openness really means in an AI context.

Another important trend David highlights is the rise of Kubernetes as a critical enabler of AI workloads. While Kubernetes was originally created to manage containerized applications, Nalley notes it’s now a go-to platform for orchestrating AI training jobs at scale. “One of the primary uses we’re seeing people use Kubernetes for is to run those training workloads,” he says.

This intersection of cloud-native technologies and AI is reshaping how companies deploy and manage their models. Tools that were once domain-specific are now being reimagined to handle the complexity of large-scale, distributed AI pipelines.

Ultimately, Nalley’s message is clear: open source is both the fuel and framework of the AI era. Whether it’s through widely adopted libraries, scalable infrastructure like Kubernetes, or the principles that underpin collaboration and access, open source plays a pivotal role in ensuring AI remains open, ethical, and replicable.

For more expert conversations and AI insights, visit TFiR