Track Overview
At PyTorch Conference North America 2026, tracks are designed around the most active areas of the PyTorch ecosystem, helping attendees focus on the sessions most relevant to their interests and expertise. From foundational concepts and core framework work to training, inference, applications, kernel engineering, and responsible AI, each track highlights a distinct part of how PyTorch is built, extended, deployed, and used in the real world.
Click on a track to learn more!
APPLICATIONS
Explore how teams are using PyTorch to build novel models, production applications, business workflows, and open-source projects across the AI stack.
Who should attend?
Practitioners, product-minded engineers, research teams, technical leaders, and developers turning PyTorch work into usable products or services.
What will you learn?
Real-world implementation patterns, applied model design, project lessons, and ways PyTorch ecosystem tools support practical AI use cases.
core pytorch
Get closer to the PyTorch framework itself, including new capabilities, framework changes, developer workflows, and project direction.
Who should attend?
PyTorch users, framework contributors, library maintainers, educators, and developers who want to understand where the core project is headed.
What will you learn?
Framework updates, internals, APIs, contribution pathways, and techniques for getting more out of PyTorch in daily development.
Inference
Learn how the PyTorch ecosystem supports efficient model serving, runtime performance, deployment, and inference at production scale.
Who should attend?
Inference engineers, platform teams, applied AI developers, systems engineers, and teams deploying models into latency- or cost-sensitive environments.
What will you learn?
Serving architectures, performance tuning, inference libraries, deployment tradeoffs, and approaches for making PyTorch models faster and easier to operate.
introduction
Learn about areas of the PyTorch library, PyTorch ecosystem or AI stack that you are not familiar with.
Who should attend?
All attendees, from students, first time attendees all the way to seasoned AI engineers that want to learn about new areas.
What will you learn?
Foundational concepts, common workflows, project orientation, and approachable ways to begin building with the PyTorch ecosystem.
kernel engineering
Go deep on compilers, optimization, custom kernels, domain-specific languages, and the lower-level systems work behind high-performance PyTorch.
Who should attend?
Systems engineers, compiler developers, performance specialists, hardware-focused teams, and experienced PyTorch developers optimizing model execution.
What will you learn?
Compiler and kernel authoring approaches, performance debugging, accelerator-aware optimization, and techniques for extending PyTorch closer to the hardware.
responsible ai
Examine the practices that make AI systems more trustworthy, including ethics, governance, security, sandboxing, and privacy across the PyTorch ecosystem.
Who should attend?
ML practitioners, security and privacy engineers, researchers, governance leads, platform teams, and decision-makers responsible for deploying AI responsibly.
What will you learn?
Risk-aware development practices, governance patterns, privacy and security considerations, and practical approaches for building safer PyTorch-powered systems.
Training
Dive into the tools, techniques, and libraries that help teams train models more effectively with PyTorch.
Who should attend?
ML engineers, researchers, data scientists, infrastructure teams, and anyone responsible for model development and experimentation.
What will you learn?
Training workflows, optimization strategies, scaling techniques, library updates, and practical guidance for improving training efficiency and reliability.
