Full Stack Engineer

In one line
We’re looking for a senior engineer who builds software by directing AI agents rather than typing code, and who can lead a production SaaS platform that way.
 

About the project
The project is a SaaS platform for culture-first hiring. It matches candidates on skills and cultural fit using psychometric profiling, a matching engine and AI-assisted assessment. The product is built and about to launch with its first pilot customers.
 

How we build: fully AI-native
The project has been built from day one with Claude Code as the main development tool. Agents write most of the code. Humans set direction, define the rules, review the output and own the result.

Concretely, the codebase has:

• CLAUDE.md files and binding playbooks (security, frontend architecture, styling) that agents must follow
• About 20 custom Claude Code skills for planning features, backend and frontend development, testing, security review, CI, dependency audits and more
• Hooks and guardrails, for example blocking agent commits to protected branches
• Parallel agents running in separate git worktrees
• MCP integrations: Playwright for visual and end-to-end testing, Jira (the roadmap syncs to Jira through Claude), Claude Design, and an accessibility checker
   

The new hire won’t just use this setup. They’ll own it, improve it, and decide how future engineers and agents work together.
 

The role
This a lead technical hire. The person drives the platform from pilot to production and owns its technical quality. They’ll work closely with the team, who also builds with Claude Code daily. We want someone who takes real ownership and pushes back, while the team keeps the final say on product direction.
 

What they’ll own

• The AI-native workflow: skills, playbooks, hooks and agent orchestration, and making the agents produce better output over time
• Judgment over agent output: reviewing AI-written code at scale, catching what drifted, and blocking what shouldn’t be merged
• Production: taking the platform live and keeping it running (deploys, monitoring, incidents)
• Security and GDPR-safe handling of sensitive candidate data
• LLM features in the product: evals, consistency, cost, prompt-injection defence and fallbacks
• Pilot feedback: turning it into scoped work
   

Must-haves

• Builds with Claude Code or a comparable agentic tool as their primary way of working, every day, for at least 6 months
• Has written their own CLAUDE.md files, skills, subagents or MCP setups, and can show them
• Can explain where agents fail and how they guard against it (tests, review, constraints, isolation)
• 7+ years in web development, with senior or lead responsibility for a SaaS product in production
• Strong TypeScript, React and Node; solid PostgreSQL. Agents write the code, but the lead has to know when it’s wrong.
• Has shipped LLM features to real users
• Pragmatic, with business sense and a startup mindset
• Excellent English and clear async communication
   

Not a fit

• “AI experience” means Copilot autocomplete or pasting into ChatGPT
• Uses AI only for boilerplate and writes the “real” code by hand
• Accepts agent output without reading the diff or running it
   

Nice to have
HR tech, recruitment or ATS experience; GDPR or EU AI Act exposure; security hardening; BullMQ, Prisma, Railway or Vercel; experience mentoring or leading.
 

Stack
React 19, Vite, Fastify 5, Prisma, PostgreSQL, Redis/BullMQ, TypeScript, Vitest, Playwright, Vercel, Railway, Anthropic models via the Vercel AI SDK.
 

Setup

• Fully remote
• Full-time and long-term
• Contract: B2B only. The candidate must have their own registered business and invoice us directly every month.
• No employment contract, and no invoicing through an agency or outsourcing company