Agentic AI Architect in London

Job Description:

Need to have experience in handling big teams.

You will lead a small, high-output agentic engineering pod migrating AWS Lambda applications to Azure.

Inspired by OpenAI’s Harness Engineering model, your primary job is not writing code — it is designing

the harness: the scaffolding, constraints, and feedback loops that let Claude Code agents do reliable

work. You own the claude_code_forge scaffolding, define CLAUDE.md specs, enforce architectural

invariants, and build the CI/CD guard-rails that keep agent-generated code coherent. You are player-

coach: 70% building harness, 30% guiding the pod.

Key Responsibilities :

• Own and evolve the claude_code_forge scaffolding — repo structure, CLAUDE.md templates,

creating agents, MCP tool configs, and custom slash commands

• Decompose the AWS estate into depth-first migration work-packages with clear acceptance

criteria agents can validate

• Enforce architecture mechanically (structural tests, linting rules, dependency-layer checks) rather

than through documentation alone

• Run weekly entropy-management sweeps using background Claude Code tasks to maintain

codebase coherence

• Mentor engineers on prompt-driven TDD, multi-agent orchestration, and Claude Desktop /

Cowork workflows

• Report migration velocity, agent utilisation, and risk to stakeholders

You Bring

• 6-8 years software engineering; 2+ years leading delivery teams

• Deep AWS and Azure serverless experience (Lambda Functions, DynamoDB Cosmos DB,

Step Functions Durable Functions)

• Hands-on Claude Code, Claude Desktop, and Cowork proficiency

• Track record of building developer tooling, CI/CD pipelines, or platform engineering