Description

Build end-to-end architecture for an enterprise agentic AI platform, leading technical decisions and defining reference architectures across squads.

This role is remote.

Responsibilities

• Design multi-agent systems using patterns like planner/executor/critic and swarm orchestration with frameworks such as Claude Agent SDK and Lang Graph.
• Define the Model Context Protocol (MCP) tool catalog, schema, server topology, and governance for tool exposure and authentication.
• Architect hybrid RAG solutions using vector and lexical search (pgvector, Postgres, OpenSearch) with chunking, re-ranking, and citation strategies.
• Establish patterns for memory management, state persistence, human-in-the-loop gates, and deterministic checkpoints.
• Drive security and compliance, including prompt-injection defenses, secret hygiene, PII handling, and model risk management.

Required Skills

• 5+ years of experience in AI/ML architecture or deep LLM application engineering.
• Hands-on expertise with at least two agentic frameworks (Claude Agent SDK, Lang Graph, AutoGen).
• Strong cloud architecture skills on AWS (Bedrock, Lambda, ECS/EKS, S3, SQS/Event Bridge).
• Deep understanding of LLM engineering: tool/function calling, structured outputs, prompt engineering, context window management, and caching.
• Experience with RAG at scale using pgvector, OpenSearch, or Pinecone, including embedding model selection and hybrid retrieval.
• Proficiency in MCP server/client design and tool integration patterns.
• Ability to define evaluation strategies, including golden sets, LLM-as-judge, and observability tools like LangSmith or Lang Fuse.
• Experience authoring ADRs, HLDs, and mentoring developers.

Preferred Skills

• Experience with model routing and cost/latency budgeting across multiple models (Opus, Sonnet, Haiku).
• Knowledge of data residency requirements and deterministic checkpointing systems.

Key Skills

Education

Any Graduate