Multi-Agent Systems Lead

A distributed intelligence architect who designs, orchestrates, and governs multi-agent ecosystems where coordination reliability, cost predictability, and auditability matter as much as emergent capability — treating agent autonomy as a privilege earned through demonstrated reliability, not a default.

- The GOVERN Multi-Agent Methodology — 6-pillar framework from problem decomposition to staged release confidence

- Agent card templates specifying identity, capabilities, contracts, and failure behaviors with typed handoff schemas

- Topology selection frameworks comparing hub-and-spoke, peer-to-peer, blackboard, and hierarchical patterns with trade-off analysis

- Failure mode taxonomy cataloging cascade failures, infinite loops, hallucination propagation, deadlocks in multi-agent contexts

- Observability pipeline design with distributed tracing, decision audit logs, and anomaly detection across agent boundaries

- Guardrail architecture with layered constraint enforcement at agent, interaction, and system levels

- Staged deployment patterns: shadow evaluation, canary rollout, regression testing, progressive autonomy expansion

- Technology stack guidance: LangGraph, CrewAI, AutoGen, LangSmith, OpenTelemetry, Temporal for orchestration and observability

- Post-incident analysis protocols with causal attribution across agent chains and cross-team knowledge normalization

Drop into Claude, ChatGPT, Cursor, or any AI tool. Bring your real multi-agent problem — a coordination bottleneck, a cost runaway, an unexplainable hallucination cascade, a topology redesign decision. It thinks like a distributed systems engineer who's shipped multi-agent systems from prototype to production governance.

Bundles or prompts related to AI orchestration and agentic architecture.