AI Engineer World's Fair 2026 — Schedule

Modern LLM applications demand infrastructure that handles massive scale reliably. We'll explore distributed serving patterns, batching strategies, and optimization techniques that work in production. Through real examples from systems processing billions of tokens daily, you'll learn how to architect inference platforms that maintain low latency while maximizing throughput. We'll dive into containerization, load balancing, and graceful degradation patterns. By the end, you'll understand the key tradeoffs in building production-grade LLM infrastructure and how to make architectural decisions that balance cost, latency, and reliability. Bring your current p99 and your monthly inference bill; the capacity model we walk through is the one that took ours from 940ms to 210ms without adding a single GPU.

Speakers: Sam SpeakerDr. Elena Rodriguez

Get hands-on experience building autonomous agent systems from scratch. We'll start with agent basics: state management, decision loops, and tool integration. You'll build a working agent that can use multiple tools and handle errors gracefully. We'll explore different agent architectures: reactive agents, planning agents, and multi-step reasoning patterns. Participants will code along with pre-configured environments. We'll discuss common failure modes and how to debug agent behavior. By the end, you'll have a solid foundation for building agents and know where to dive deeper for production systems. Come with a laptop and a Python 3.11 environment. You will leave with a running agent, a trace of every tool call it made, and a short list of the mistakes I made first so you can skip them.

Speakers: Sam SpeakerYuki Tanaka

Standard benchmarks often miss the failures that matter in production. This talk explores building comprehensive evaluation frameworks that actually predict real-world performance. We'll discuss what makes evals brittle versus robust. You'll learn techniques for designing evals that are rigorous without becoming expensive or gaming-prone. We'll cover automated evaluation systems, human-in-the-loop approaches, and continuous monitoring. Practical examples from production systems show how different evaluation strategies catch different failure classes. Walk away with actionable patterns for building better evals. Fair warning: about half of this talk is failures — three eval suites we shipped that stayed green while the model got measurably worse, and what we changed to make them able to fail.

Speakers: Rina OkaforMarcus Chen

You don't need billion-dollar budgets to build impactful AI tools. This talk explores how solo founders and small teams are shipping competitive AI products through smart architecture choices. We'll discuss cost optimization strategies, efficient model selection, and architectural patterns that maximize impact per dollar spent. Real examples show how open-source foundations and clever engineering solve problems previously requiring massive resources. Learn about on-device inference, caching strategies, and using models appropriately. Discover how constraint can drive innovation and how to build sustainable AI products as a bootstrapped founder. Aimed at solo builders and two-person teams. Every number in this talk comes off a real invoice, and the whole stack runs for less than the cost of a team lunch.

Speakers: Jamie Whitlock