Dima Dzhulgakov

Co-Founder & CTO, Fireworks AI

Dima is a co-founder of Fireworks AI, the inference and reinforcement learning platform that makes it simple, fast, and cost-effective to build and scale AI products. Fireworks AI powers leading AI startups and enterprises like Cursor, Doordash, Notion and Uber. Dima is one of PyTorch’s core maintainers. Previously, he helped to bring PyTorch from a research framework to numerous production applications across Meta’s AI use cases and broader industry.

Dima Dzhulgakov

Sessions / 2026 / 1 talk

  • Closed frontier models get the headlines, but the agentic revolution is being built on something more flexible and more interesting: a hybrid stack of open-weight models, intelligent routing, and high-throughput inference infrastructure. In this panel, founders building the open AI ecosystem — from fine-tuned model deployment to multi-model orchestration to optimized routing and serving — get into what "open" actually means in 2026, and where it falls short. We'll dig into the real engineering tradeoffs: when open models win in agentic pipelines, when they don't, how routing and inference choices shape latency and cost at scale, and what it takes to build reliable agents in this open ecosystem as always-on agents become an increasingly core part of agentic engineering.

Ready to take this stage?

The next edition is programmed by practitioners. Tell us what you built and what you learned.

Apply to be a speaker