Engineering AI Agentic Systems: Decisions, Tradeoffs, and Production Reality

Topic: Engineering AI Agentic Systems: Decisions, Tradeoffs, and Production Reality

Agenda:
6:00 PM – Networking and refreshments
6:30 PM – Presentation commences
7:15 PM – Q&A
7:30 PM – Networking and refreshments
8:00 PM – Event concludes

Abstract:

Everyone can get an agent working in a weekend. Shipping one to production is a different problem.
This talk is a systems engineering lens on agentic AI, starting from the moment a single LLM call becomes a decision loop through the architectural tradeoffs that determine whether it survives contact with reality.

We'll cover the control spectrum between deterministic workflows and autonomous agent loops, multi-agent coordination, state management, and what a real control plane looks like beyond rate limiting and logging.We'll also cover the failure modes that don't appear in demos: runaway costs, missing stopping conditions, and observability gaps.

You'll leave with a framework that treats human-in-the-loop design, model selection, and control plane engineering as first-class architectural decisions, because agentic systems are not prompt engineering problems. They are systems engineering problems.