Alex Albert Head of Developer Relations, Anthropic
Leads developer relations at Anthropic, working with the engineers building on Claude.
Closing panel: prototype to production
Jo Kristian Bergum Chief Scientist, Vespa.ai
Works on search and retrieval at scale, and on the parts of RAG that break once the corpus grows.
Grounding copilots without a vector db meltdown
Will Brown Research Lead, Prime Intellect
Researches reinforcement learning for language models and decentralized training.
Workshop: build a coding agent fleet
Charles Frye Developer Advocate, Modal
Teaches people to build and serve models, and works on the infrastructure side of getting GPUs to do useful work reliably.
Latency budgets for multimodal agents
Sarah Guo Founder, Conviction
Founder of Conviction, investing in AI-native companies. Hosts the No Priors podcast.
Closing panel: prototype to production
Dex Horthy Founder, HumanLayer
Builds HumanLayer, which puts human approval in the loop for agents that take real actions.
Building reliable agent loops
HH Hamel Husain ML Engineer, Independent
Consults on evaluation and fine-tuning, and writes about why most teams' eval suites do not measure what they think they measure.
Evals that catch regressions before users do
Sayash Kapoor PhD Researcher, Princeton
Co-author of AI Snake Oil. Studies where AI claims outrun AI evidence, and what reproducible evaluation actually requires.
Regression detection at model-update speed
Andrej Karpathy Co-founder, OpenAI
Founding member of OpenAI and former Director of AI at Tesla. Writes and teaches widely on neural networks and the craft of building with them.
Opening keynote: Agents that ship
Omar Khattab Assistant Professor, MIT
Creator of DSPy. Researches how to program language models rather than prompt them, and how to make those programs hold up.
Jerry Liu Co-founder & CEO, LlamaIndex
Builds LlamaIndex, the data framework for LLM applications. Works on retrieval, agents, and getting private data into production systems.
Hands-on: malleable software with agents
Taming 40-Minute CI: Incremental Builds at Monorepo Scale
Philipp Schmid Staff Engineer, AI DevRel, Google DeepMind
Works on making open models practical to deploy, and on the tooling that surrounds them.
Substituting models safely
Simon Willison Creator of Datasette, Independent
Co-created Django and now builds Datasette, an open-source tool for exploring and publishing data. Writes daily about what LLMs can and cannot do.
Closing panel: prototype to production
Eugene Yan Principal Applied Scientist, Amazon
Works on recommendation and search systems at scale, and writes extensively on evaluation and applied machine learning practice.
Serving embeddings for 40k QPS
BZ Barry Zhang Applied AI, Anthropic
Works with teams putting Claude into production, and on what separates a demo agent from a dependable one.
Live-coding a conference copilot
