Agentic AI in Production: Scaling, Observability, and Human-in-the-Loop Orchestration: MLOps pipelines, scaling strategies, observability, human overs

By Marcus Endo

Operational guide for taking agentic prototypes to robust, scalable production services. Covers session and state checkpointing, horizontal scaling, autoscaling heuristics, model caching and selection, cost-control (quantization, distillation), observability tailored to agent workflows (traceability of tool calls, prompt histories), human-in-the-loop escalation patterns, and governance for auditability and explainability.
Who this book is forProduction ML engineers and platform teams deploying agents at scale.SREs and DevOps engineers managing latency, throughput, and uptime.Product owners who need SLAs, auditability, and governance around agents.Teams adopting human-in-the-loop controls for quality and safety.What the reader will learnProduction architectures for long-lived agent sessions and stateful orchestration.Cost-performance tradeoffs: quantization, model distillation, and routing strategies.Observability design: metrics, logging, traces for tool calls and prompt histories.Human-in-the-loop and escalation mechanics to ensure safety and quality.Versioning, canarying, governance, and privacy-preserving deployment patterns.Scaling strategies for multi-tenant agent platforms and SLA design.

• Format: Paperback
• Language: English
• ISBN: B0FQTLZNYG
• ISBN13: 9798264891052
• Release Date: September 2025
• Publisher: Independently Published
• Length: 334 Pages
• Weight: 1.28 lbs.
• Dimensions: 0.7" x 7.0" x 10.0"