Six Pathways Through the Talks
Finding your way through three days of world-class Python content
PyCon US 2026 runs May 13–19 in Long Beach, California, and with over 100 talks across five rooms over three days, the schedule can feel like a lot to navigate. The good news: whether you came to go deep on Python performance, level up your security knowledge, get practical Python insights for agentic AI, or finally understand what all the async fuss is about, there's a clear path through the content that's built for you. Register now to get in on the full experience.
We mapped six attendee pathways through the full talks schedule with a bonus tutorial to pair with it, each one a curated sequence of sessions that focuses on a core Python topic. Think of them less as tracks and more as through-lines. Pick the one that matches where you are and what you want to walk away with to integrate into your work.
Python Performance: From Memory to Metal
If you want to understand why your Python is slow and what to actually do about it, this is your path. It runs across all three days and takes you from memory profiling fundamentals all the way to CPython internals with one of the core developers who is actually changing the way the runtime works.
Friday
Goutam Tiwari's I Accidentally Built a Monitoring System While Trying to Debug a Memory Leak: a grounded, story-driven entry point into how memory and profiling interact in real systems.
Wenxin Jiang and Jian Yin's Breaking the Speed Limit: Fast Statistical Models with Python 3.14, Numba, and JAX gives you hands-on acceleration tools you can take home and use immediately.
Thomas Wouters, a CPython core developer and Steering Council member, delivers Free-threaded Python: past, present and future, the definitive account of GIL removal from the people doing the work.
Matthew Johnson's Lock-Free Multi-Core Performance with Behavior-Oriented Concurrency
Bruce Eckel's Demystifying the GIL to close out Friday with a complete mental model of where Python concurrency has been and where it's going.
Saturday
Larry Hastings' Conquer multithreaded Python with Blanket, practical multithreading tooling that builds directly on Friday's foundation
Jukka Lehtosalo on Making Python Faster with Free Threading and Mypyc
Yineng Zhang's High-Performance LLM Inference in Pure Python with PyTorch Custom Ops, which applies everything you've learned to one of the most demanding production workloads in the industry right now.
Sunday
Mark Shannon's Memory management in CPython, fast or slow? is as close to the source as it gets: a look at the engine underneath all your performance gains, from a core contributor who has spent years making it faster.
Pair it with a tutorial: Start the week with Arthur Pastel and Adrien Cacciaguerra's Wednesday tutorial Python Performance Lab: Sharpening Your Instincts. It's a hands-on lab designed to build the kind of performance intuition that makes everything in this pathway land harder.
Debugging and Observability: Finding What's Wrong and Why
This pathway is for engineers who spend too much time in production fires and want better tools for preventing and diagnosing them. It moves from memory leak storytelling through the brand new profiling and debugging interfaces, landing in Python 3.14 and 3.15.
Friday
Goutam Tiwari's memory leak talk is your opening again; it's the most relatable entry point in the schedule for anyone who has ever stared at a climbing memory graph, wondering what went wrong.
Puneet Khushwani's Demystifying Python's Generational Garbage Collector gives you a clear foundation
Anshul Jannumahanti's Debugging Python in Production: Practical Techniques Beyond Print Statements gives you the toolkit to act on it.
Saturday
Pablo Galindo Salgado (keynote speaker and Steering Council member) and Laszlo Kiss Kollar present Tachyon: Python 3.15's sampling profiler is faster than your code — brand new profiling infrastructure in the language itself, from one of the people who built it.
Running concurrently, fellow Steering Council member Savannah Ostrowski's The art of live process manipulation with Python 3.14's zero-overhead debugging interface demonstrates how to inspect live Python processes with no performance penalty at all. These two together represent a genuine step change in what's possible for Python observability.
Pair it with a tutorial: Catherine Nelson and Robert Masson's Thursday tutorial Going from Notebooks to Production Code is a natural warm-up, it covers the gap between exploratory code and production systems, which is exactly where most debugging pain lives.
Concurrency and Async: Making Python Do More at Once
The concurrency story in Python is changing faster than it has in years. This pathway traces the thread from hardware-level parallelism through the GIL removal to practical async patterns for the systems people are actually building in 2026.
Friday
Benjamin Glick's GPU Communications for Python: hardware context before software patterns.
Thomas Wouters on Free-threaded Python gives you the foundational GIL story.
Aditya Mehra's Don't Block the Loop: Python Async Patterns for AI Agents provides a real-world application of event loop patterns in production systems.
Matthew Johnson's Lock-Free Multi-Core Performance
Bruce Eckel's Demystifying the GIL rounds out Friday
Saturday
Conquer multithreaded Python with Blanket by Larry Hastings brings it home with practical tooling for production multithreaded Python.
Pair it with a tutorial: Trey Hunner's Wednesday tutorial Lazy Looping in Practice: Building and Using Generators and Iterators is a perfect primer. Generators and iterators are the building blocks of Python's async model, and Hunner is one of the best teachers in the community at making these concepts click.
AI and Machine Learning: From Inference to Agents
The dedicated Future of AI with Python track runs all day Friday, May 15th, and it's one of the strongest single-day lineups in the schedule. This pathway threads the AI content across the full conference, from hardware fundamentals to production-grade inference.
Friday
Benjamin Glick's GPU Communications for Python sets the hardware context.
Aayush Kumar JVS's Running Large Language Models on Laptops: Practical Quantization Techniques in Python is one of the most immediately practical talks in the schedule, if you have ever wanted to run a model locally and not known where to start, this is your session.
Aditya Mehra's Don't Block the Loop covers the async foundations that make reliable agents possible.
Santosh Appachu Devanira Poovaiah's What Python Developers Need to Know About Hardware demystifies GPU memory and execution models in a way that's genuinely useful for anyone writing inference code, and
Camila Hinojosa Añez and Elizabeth Fuentes close out Friday's AI track with How to Build Your First Real-Time Voice Agent in Python.
Saturday
Yineng Zhang on High-Performance LLM Inference in Pure Python: production-grade optimization for teams shipping at scale.
Pair it with a tutorial: Two tutorials are worth your attention here. Pamela Fox's Wednesday Build Your First MCP Server in Python is the fastest way to understand how agentic systems actually work under the hood — MCP is quickly becoming the standard way to give AI agents access to tools and data. And Isabel Michel's Wednesday Implementing RAG in Python: Build a Retrieval-Augmented Generation System gives you the hands-on foundation underneath most modern LLM applications.
Security: A Full Day Worth Taking Seriously
Saturday, May 16th, is the first-ever dedicated security track at PyCon US, and if security is anywhere near your professional concerns, you should plan to spend most of Saturday in Room 103ABC. Eleven experts. One room. A full day.
Saturday
The day opens with Ian's FastAPI Security Patterns: OAuth 2.0, JWTs, and API Keys Done Right — the fundamentals every Python web developer should have down.
PSF’s own PyPI Safety & Security Engineer Mike Fiedler's Anatomy of a Phishing Campaign flips the lens and gives you the attacker's perspective before you go back to building defenses.
Tristan McKinnon's Zero Trust in 200ms covers modern identity architecture, and
Emma Smith's Rust for CPython looks at the language-level safety improvements coming to CPython itself.
Sanchit Sahay and Abhishek Reddypalle on SBOMs for Python Builds
Andrew Nesbitt on GitHub Actions Security,
Hala Ali and Andrew Case on Post-Incident Runtime SBOM Generation from Python Memory
Shelby Cunningham and Madison Ficorilli are closing with Breaking Bad (Packages): Why Traditional Vulnerability Tracking Fails Supply Chain Attacks. If you've been meaning to get serious about supply chain security, this is the day to do it.
Pair it with a tutorial: Paul Zuradzki's Wednesday tutorial, Practical Software Testing with Python is a strong complement: the discipline of writing tests and the discipline of writing secure code overlap more than most developers realize, and this tutorial gives you the testing foundation that makes security practices easier to implement and verify.
New to Python and Packaging: A First-Timer's Path Through the Conference
Not every pathway is about going deep. This one is for attendees who are newer to Python or who want to level up on tooling, packaging, and writing code that other people can actually use. It runs gently across all three days and ends with a satisfying arc.
Friday
Russell Keith-Magee's How to give your Python code to someone else, the distribution problem from first principles, from one of the most thoughtful voices in the Python community on the topic.
Zanie Blue's Peeking under the hood of uv run covers the modern tooling that's quickly becoming the standard
Trey Hunner's pathlib: why and how to use it is the kind of practical skills upgrade most developers underestimate until they've seen it.
Saturday
Justin Lee's Python for Humans — Designing Python Code Like a User Interface, which will change how you think about writing APIs and interfaces for other developers.
Mario Munoz's Create a Python Package: From Zero to Hero puts the whole packaging arc together in one session.
Sunday
Rafael Mendes de Jesus on From notebooks to scripts: turning one-off analysis into reusable Python code: the graduation moment from exploratory to production-ready.
Pair it with a tutorial: Mason Egger's Thursday tutorial, Writing Pythonic Code: Features That Make Python Powerful, is the ideal warm-up for this entire pathway. It covers the idioms and language features that separate code that works from code that feels like Python, which is exactly the mindset the rest of this track builds on. Or if you are just getting started with no experience at all, try Python for Absolute Beginners. If you've started and stopped learning to code before, or never got around to starting at all, sign up for this tutorial and start PyCon on a strong step.
However you come to PyCon US 2026, there's a path through the schedule built for you. The full talks schedule is at us.pycon.org/2026/schedule/talks, the full tutorials schedule is at https://us.pycon.org/2026/schedule/tutorials/, and registration is open now.
We'll see you in Long Beach.
PyCon US 2026 takes place May 13–19 in Long Beach, California. Talks run Friday, May 15th, through Sunday, May 17th.
Comments