AI Product Engineer (Fullstack)

We believe talent is everywhere. The chance to live from it isn't. Millions of creators are talented and invisible at the same time. The value they create flows to middlemen, while the brands who want to reach them are stuck in a non-transparent, fragmented market. So we built the tools to fix both sides at once. Creators turn an account into an income: join a campaign in minutes, make something they're proud of, and get paid fairly for creativity, not follower count. Brands brief once, activate hundreds of creators, and see content live within 24 hours. Next year, 10,000 creators will make a living from WePush. When a creator earns, a brand grows. When a brand grows, more creators earn. WePush is the fastest-growing creator marketing platform in Europe, with over 150,000 creators and 500,000+ collaborations powering brands like L'Oréal, ABOUT YOU, Zalando, and Sony Music. Two numbers matter to us: how many people make a living through WePush, and how many millions of people brands reach through them. **Technical Overview: **We run on a real distributed system: Angular and React frontends, a native iOS app, a Node/TypeScript application server handling booking and matching logic, a fleet of independently deployable backend services on Kubernetes, a real-time layer, a ledger that moves actual money, and a small army of scrapers. We're looking for a Fullstack AI Product Engineer who is comfortable moving across all of those layers and who uses AI agent tools as genuine leverage. We use these tools daily to read unfamiliar services, ship changes, and move fast. Tasks What you'll actually work on: A native iOS app: SwiftUI, rewritten natively Web Frontend: our marketplace apps are built in Angular, and newer internal tooling is React. An application server layer: a Node/TypeScript service sitting between the frontend and the backend fleet, handling booking, scheduling/matching logic, invoicing, and PDF generation. A polyglot microservices fleet, with a shrinking Hack/HHVM legacy core and a growing set of services in Node.js and Go, each independently versioned, containerized, and deployed via Helm to Kubernetes A real CI/CD pipeline: GitLab CI building images with Kaniko, tag-based semver releases (nightly unstable builds, stable releases promoted through dev to test to a manually gated prod), and cross-service pipeline triggers when a shared dependency publishes Scheduled workloads at scale: a couple dozen CronJobs per environment (reminders, payout checks, stat rollups, token refreshes, media polling), each independently schedulable, tunable, and killable without redeploying the whole service Money movement: a ledger service backing real balances and payouts A scraper fleet: services that scrape platforms which actively fight back (rate limits, bot walls, reshaped responses), running on their own schedule with their own failure modes A newer AWS/serverless edge: internal tooling and some integrations are moving onto Lambda/API Gateway/DynamoDB. This is where a lot of new, greenfield work is happening What "AI-driven builder" means here: You use AI tools openly, as your default way of working, to read and understand an unfamiliar service fast, not just to generate new code. Half the job is "what does this cron job actually do and what breaks if I touch it," and AI is how you get that answer in minutes instead of a day You verify before you trust: you'd catch a plausible but wrong AI suggestion before it ships You can operate across languages and layers without needing to be a specialist in all of them, including ones you don't know yet You respect the promotion pipeline. You don't route around dev to test to prod because it's slower The stack you'll touch: Angular and React on the web frontend, native iOS in SwiftUI, a Node/TypeScript application server, Node.js and Go for newer backend services, a legacy Hack/HHVM core we're winding down, Kubernetes, Helm, GitLab CI, Docker, a Formance-based ledger, WebSockets, cron-d