AI Agent Orchestration for Enterprise Arch with AWS Bedrock, Flyte & LangGraph
summary
Built an enterprise AI orchestration stack that combines AWS Bedrock (Claude 3) for inference, LangGraph for multi‑agent state and execution plans and Flyte for parallel orchestration, delivering an architecture copilot grounded in a domain knowledge graph of 200k+ nodes. The system runs ~56 parallel recommendation tasks, issues 500+ LLM calls per recommendation, produces 50–100 recommendations per run, enforces token/time budgets, achieves 100% task completion with multi‑level retries, and adds OpenTelemetry tracing for full observability.
AI-written summary · about
chapters (8)
- 0:00Enterprise AI orchestration intro
- 3:27Reasoning system for enterprise arch
- 8:03AWS Bedrock foundation model layer
- 13:25LangGraph agent execution engine
- 19:03Flyte parallel task orchestration
- 24:24Production results and recommendations
- 26:12Lessons from multi-agent systems
- 28:40Q&A
related talks
26mGrounded Reasoning Systems for Cloud Architecture
17mAI TechOps: Data and Infra for AI Foundation to Scale and Manage 1000s of Cohesive Agents
43mBeyond Chatbots: How Amazon Bedrock Agents Drive Enterprise AI
28mBuilding Agentic AI applications on AWS
31mAgentic AI architecture patterns
22mAI Agents for Multiprovider Cloud Edge Continuum
43mArchitecting the Future: Modern Solution Design with Agentic AI
27mAI-Powered ETL Pipeline Orchestration: Multi-Agent Systems in the Era of Generative AI