Senior Principal Architect Agentic AI & Digital Platforms

Our client is looking for a Senior Principal Architect Agentic AI & Digital Platforms to serve as the onshore technical anchor for an Agentic AI initiative across customer-facing digital platforms. The ideal candidate will translate business intent into scalable governed production-ready systems and hold the architectural line across delivery teams. The right person codes reviews and debates trade-offs in the same week they brief a VP.

Key Responsibilities:

• Agentic Architecture - Design & govern end-to-end agentic architecture for apps loyalty payments and commerce workflows.
  
• Multi-Agent Orchestration - Orchestrator/subagent hierarchies tool calling stateful graph workflows agent memory layers.
  
• MCP Standards - Establish Model Context Protocol standards for governed tool and enterprise service exposure.
  
• Runtime Guardrails - Spend controls fraud hooks human-in-the-loop checkpoints and full auditability trails.
  
• Retail Systems Integration - Lead AI agent integration with catalog inventory pricing promotions loyalty and checkout.
  
• LLMOps Standards - Prompt versioning RAG governance evaluation harnesses observability (latency cost drift safety).
  

Requirements

Must-have (non-negotiable):

• 1218 yrs enterprise/platform architecture
  
• 34 yrs AI/ML or GenAI in production
  
• Shipped LLM-based agent systems (not POC)
  
• Hands-on: LangGraph / LangChain / Semantic Kernel / CrewAI / AutoGen
  
• ReAct multi-agent hand offs RAG MCP knowledge
  
• Architected against hallucination tool misuse cost runaway
  
• Python proficiency reference implementations
  
• Translates tech business value for VPs
  

Preferred:

• Direct retail / commerce domain exposure
  
• LLMOps depth (LangSmith / Langfuse)
  
• Onshore/offshore delivery leadership
  
• Knowledge graphs as grounding layers
  

Required Skills:

>1218 yrs enterprise/platform architecture 34 yrs AI/ML or GenAI in production Shipped LLM-based agent systems (not POC) Hands-on: LangGraph / LangChain / Semantic Kernel / CrewAI / AutoGen ReAct multi-agent handoffs RAG MCP knowledge Architected against hallucination tool misuse cost runaway Python proficiency reference implementations Translates tech business value for VPs