Senior Full-Stack Engineer to Ship AI Features in a Live SaaS

We run an established B2B SaaS with real customers on a multi-tenant platform. Our AI roadmap is growing faster than our in-house team can keep up with, so I'm looking for one senior engineer who can genuinely own AI features end to end — understand the user problem, design the flow, build the frontend, write the backend, wire up the model, test it, and ship it. We've already shipped an AI assistant and RAG-based search, so you won't be starting from zero. The infrastructure, patterns, and conventions exist. What's next on the roadmap: • An AI-powered onboarding flow configures a new workspace from a plain-English description of the customer's business. • Inline AI actions across our existing forms — rewrite, extract, and classify, right where users are typing. • An insights page that sits on top of our analytics data and turns it into readable AI-generated narratives. How we'll start I don't want to hand a stranger the whole roadmap on day one, and honestly you probably don't want that either. So we start small: one paid, clearly scoped task — a single inline AI action (summarize + classify) on one of our existing forms, built end to end from the UI button to the Claude call to the saved result. You'll get Slack and GitHub access, work on a feature branch, and I'll personally review your PR. A detailed technical spec for this first task is ready and I'll share it as soon as we start talking, so you'll know exactly what you're walking into. What I'm looking for The kind of engineer who takes a one-paragraph feature brief and comes back with a shipped PR, having asked sensible questions along the way. Someone who cares about how AI actually feels in a UI — streaming instead of spinners, optimistic updates, graceful failures instead of broken pages. And someone who writes code the rest of the team can maintain after the contract ends. You should be comfortable with • Next.js / React / TypeScript at a senior level, including complex state management and streaming UI. • Node.js and Python backend development — REST and GraphQL. • Claude API and OpenAI API: prompting, tool calling, structured outputs, streaming. • PostgreSQL and working safely inside a mature multi-tenant codebase you didn't write. • Owning a feature end to end with minimal supervision — you flag blockers early instead of going quiet. Nice to have (not required) Vercel AI SDK, Tailwind, WebSockets, LangChain, or prior experience shipping B2B SaaS products. How to apply Please skip the generic copy-paste pitch, I can tell, and it goes straight to the archive. Start your proposal with the word “Harbor” so I know you actually read this far. Then, in a few lines, tell me about one AI feature you built completely on your own — UI to backend to model — and what you would improve about it today. Links to live work beat certificates every time. Screening questions • Tell us about an AI feature you built completely on your own: frontend, backend, and model integration. What did you ship, and what would you improve now? • Our features must feel instant even when the model takes seconds. What patterns do you use (streaming, optimistic UI, background jobs) to keep AI UX fast — and where have you actually applied them?