For experienced developers and engineering leaders. Three live engineering deep-dives: harnesses, agentic search and real-time voice. Each session covers design, trade-offs and performance, tested in production. You learn the engineering decisions, not the features of a tool. 3 hours, online, max 15 people.
The harness is the loop, the tools and the checks around a model. It is the shared layer under search and voice. It decides what an agent costs and how much you can trust it.
Build agentic search with a coding agent. Set up the data and the evals first. Compare grep ranking, BM25 and vector search on e-commerce and finance datasets, and find where each one fails. Then scale to multi-hop queries across thousands of documents and long entity chains.
Build a voice agent that answers in real time. Find where latency, interruptions and echo break it. Apply the fixes that hold in production.
Select the sessions you want. We send you an email when the dates open. Each session has at most 15 engineers.