This is a blog series where we're asking each of our PyConUS 2026 keynote speakers about their journey into tech, how excited they are for PyconUS and any tips they can provide for an awesome conference experience! Here's our interview with Lin Qiao
Without giving too many spoilers, tell us what your keynote is about?
Most AI products are built on rented land. If your competitor can make the same API call, you do not have a moat. I will break down what the teams pulling ahead are doing differently, with real examples from Cursor, Notion, and Vercel, and get into the hard tradeoffs nobody talks about enough.
How did you get started in tech/Python?
My path into tech started pretty naturally. I studied STEM all through high school and undergrad, so it was always the space I gravitated toward. Python specifically came later, during my PhD, where I started using it to run experiments and support my research papers.
What do you think the most important work you've ever done is?
Co-creating PyTorch was a defining chapter, because it became the foundation for how the world does AI research. But I think the most important work is really what I’m doing now. I founded Fireworks because I spent years watching companies outside Big Tech struggle to get AI into production. They had the ambition but not the infrastructure, and we’re changing that.
Have you been to PyCon US before? What are you looking forward to?
PyTorch was built on Python, so this community is close to my heart. I am most looking forward to the hallway conversations. The best ideas come from talking to people who are deep in the work.
Any advice for first-time conference goers?
Talk to people. The sessions are recorded, but the people are only there for a few days. Go to the hallway track, sit at lunch tables where you do not know anyone, and if a talk resonated with you, go tell the speaker. That’s how the best professional relationships start.
Can you tell us about an open source or open culture project that you think not enough people know about?
I am biased, but I think the broader open model ecosystem does not get the credit it deserves. Everyone knows the big names, but there is an incredible amount of work happening in specialized open models, evaluation frameworks, and fine-tuning tooling that is quietly making AI more accessible. The pattern I keep seeing is that the most impactful open-source projects are the ones that lower the barrier for the next person to build something better. That was true for PyTorch, and it is true today for the tools that help developers go from an off-the-shelf model to something truly customized for their use case.
Comments