Product Engineer
Product Engineer – Claude Code / AI Agents
Hybrid – 1 / 2 days per week in London
Salary very much dependant on skills and experience
Do you have a background in software development / product engineering and experience in using AI Agents for producing code? Do you like to question outputs and think of the overall system context? Can you think and not just type? Please get in touch!
This is a new position in a start up environment, where there is a plan to move from an outsourced platform development model to inhouse, using Claude Code.
You won’t be writing code, but you will be constantly thinking about architecture, assessing plans, guiding AI execution and validating outputs. You will be the human in the AI Native engineering model - providing judgement calls on the AI Agents code – is it right or does it just work? Is it the right solution or just based on assumptions?
As the Product Engineer, you will steer the AI Agent through a structured methodology, applying judgment at every gate, and ensuring the output is architecturally sound, production-quality, and aligned with the product vision.
Due to the start up nature of the role you will work directly with the COO and CTO to shape the product, build new features, harden the existing platform, and establish the engineering practices that will scale as the company grows.
Core Accountabilities will include:
- Translate business objectives, customer feedback, and market opportunities into clear technical direction
- Own the translation of product intent into effective briefs that the factory can execute against
- Define and maintain the target technical architecture for the platform, ensuring every Claude Code output aligns with it
- Continuously review architectural decisions as the codebase evolves to avoid AI Drift
- Own the system design documentation that feeds Claude Code’s context
- Critically assess every plan Claude Code generates before approving execution
- Guide Claude Code through effective prompting: precise context injection, constraint specification, and decomposition of complex work
- Monitor chain performance: where does Claude Code excel, where does it struggle, what context improves output quality - Catch AI-specific failure patterns
- Build and refine the context layer (architecture docs, coding standards, domain rules) that feeds the chain - When the chain fails, fix the chain.
You will come from a background in development / engineering and strong coding experience but with great architecture and overall system knowledge.
You will have pivoted to AI-augmented workflows
You will have experience in AI Agents – Claude Code, Cursor, Devin, or similar tools and understand their limitations as well as their strengths.