Aimed at compiler engineers who already know LLVM and MLIR. The goal is to give that audience a map of how Mojo works. We will provide an inside view of the compiler, type system, and GPU backend from the people who built them. We will also discuss Mojo's role in the LLVM ecosystem and community.
Full-day workshop
8:30 – 8:40 AM | Welcome
Denis Gurchenkov
8:40 – 9:30 AM | Mojo’s Place in the LLVM Ecosystem
Weiwei Chen, Billy Zhu, Lukas Hermann
9:30 – 10:00 AM | Mojo Compiler Architecture Overview
Billy Zhu, Weiwei Chen
10:00 – 10:30 AM | Break
10:30 – 11:00 AM | Mojo Plugin Architecture: Separating the Compiler and SDK from Hardware-Specific Components
Weiwei Chen, Manoj Rajagopalan, Connor Gray
11:00 – 11:30 AM | Value Ownership, Lifetimes, and Origins: Mojo’s Memory Safety Model
Chris Lattner
11:30 AM – 12:00 PM | Mojo’s LLVM and MLIR Backends
Kolya Panchenko
12:00 – 1:00 PM | Lunch
1:00 – 1:30 PM | MAX Graph Compiler and MLIR Dialect for Kernel Fusion
Feras Boulala
1:30 – 2:00 PM | How the Mojo Standard Library Provides a Friendly Interface over MLIR Constructs
Joe Loser, Nathan Ward
2:00 – 2:30 PM | Case Study: How Mojo Metaprogramming Abstracts Away Hardware Specificity
Kern Handa
2:30 – 3:00 PM | Mojo Roadmap
Denis Gurchenkov
3:00 – 3:30 PM | Break
3:30 – 4:00 PM | Modular Open Source Contribution Strategy
Brad Larson, Joe Loser
4:00 – 4:30 PM | Q&A
4:30 – 5:00 PM | Mojo and the LLVM Community Roundtables
5:00 PM | Close