Stephen Toub: AI in .NET - Episode 344

Stephen Toub is a Partner Software Engineer at Microsoft. Stephen has been at Microsoft for almost 24 years and has done it all. If it has to do with .NET, he's been in it. And today, he looks after all the .NET libraries making sure .NET continues to be the fastest platform on the planet. He ran the MSDN Magazine before it morphed into the docs and blogs of today. He was a leader in the concurrency and async and parallel programming developments, and now he's turned his attention to pushing the AI envelope with .NET.

Topics of Discussion:

[4:37] Stephen morphed a few of his many interests into the role he has today.

[6:04] How moments like the push for parallel computing, the open-sourcing of .NET, cross-platform support, and performance goals shaped Steven's current focus.
[7:37] Steven highlights the two major AI tracks: (1) AI-augmented tools that improve developer productivity, and (2) building AI capabilities into applications themselves.

[10:37] The future of user interfaces.

[12:43] Why "agents" are basically advanced actor-based systems empowered by large language models and tool calling, enabling reasoning and orchestration beyond simple Q&A.
[23:22] Model context protocol.

[24:08] A paradigm shift in system integration.

[30:24] How Stephen and the .NET libraries team factor out common AI abstractions for the ecosystem, so that different frameworks (Semantic Kernel, Onnx Runtime, Olama, etc.) can interoperate.

[32:15] Steven gives examples of how minimal C# code, combined with locally hosted LLMs or cloud endpoints, can solve real tasks.

Mentioned in this Episode:

Clear Measure Way

Architect Forum

Software Engineer Forum

Programming with Palermo — New Video Podcast! Email us at [email protected].

Clear Measure, Inc. (Sponsor)

.NET DevOps for Azure: A Developer's Guide to DevOps Architecture the Right Way, by Jeffrey Palermo

Stephen on GitHub

Stephen on LinkedIn

Demystifying Retrieval Augmented Generation with .NET

Suno