BUILD THE FUTURE WITH PLAY PORTFOLIO COMPANIES

Full-stack Engineer

Scenario

Software Engineering

Paris, France · Lyon, France

Posted on Jun 18, 2026

Ship features across the web app — agent chat, workflow editor, model & inference surface, billing and seat management, analytics and Canvas.
Own the AI-surface experience: streaming responses, tool-call rendering, structured output, latency budgets and graceful error recovery on long agent runs.
Improve the AI layer and the agent surface end to end — tool calls, multi-step agents, structured outputs, model routing, compaction, transcript persistence, MCP integrations and prompt caching.
Keep the frontend healthy: patch security vulnerabilities (Dependabot, GitHub advisories, SOC 2 findings) on a steady cadence, and co-own the architecture choices that keep the app fast and coherent as it grows.
Open pull requests in the cloud API, the TypeScript SDK and the MCP server whenever the right fix lives there, and coordinate cross-stack PR trains with the backend and ML teams.
Bring AI pair-programming into the team's daily flow: the prompts, agents and Claude-Code-driven workflows that make everyone faster.

Owns

The frontend, end to end and all its features.
AI infrastructure and power-user tooling: the AI layer and the agent surface.
Full-stack reach: cloud API PRs, SDK and MCP-server extensions, cross-stack coordination.

Does not own

Four to seven years in frontend or full-stack, ideally on a SaaS product with a non-trivial real-time or streaming surface.
Strong TypeScript and React or Next.js, comfortable reading Node/serverless code and confident enough to ship a Vercel Function, a SDK change or even an AWS Lambda.
Hands-on with at least one AI SDK in production — Vercel AI SDK preferred, Anthropic or OpenAI acceptable; you've shipped streaming chat, tool calls or structured output, not just demos.
A daily user of AI coding assistants (Claude Code, Cursor, Copilot) with a strong bias for action across the stack, and clear written English when proposing changes that touch other teams’ code.
Background in generative AI / image-video models.

A frontend-only mindset that refuses to open a pull request in the API or an SDK when the bug lives there.
No hands-on AI SDK experience: hasn't shipped streaming chat, tool calls or structured-output features in production.
Not a daily user of AI coding assistants; thinks in tickets rather than agents and prompts.
Treats dependency hygiene and security patching as someone else's problem.