AI Software Engineer

AI Software Engineer An unsolved problem in a multi-billion-dollar industry. A revenue-generating startup with signed commitments and paying customers. A build-from-the-front engineering role with ownership - where the hard part is the software, not the model. This is a software engineering role in an AI domain, not an AI role that happens to involve software. Our client builds agents and automation that take document-heavy business workflows and run them end to end: a document lands, an email arrives, and a reusable agentic workflow fires automatically - extracting what matters and acting on it, instead of someone re-prompting a chatbot every time. The product is live, customers are paying, and the team is small and fully distributed, so what you build, you own. The hard part here is the engineering. Production-grade backend is live Agent experiences is what they are hiring for. What You'll Be Doing • Build and own the backend the whole product runs on - services, APIs, workers, queues and event-driven workflows in Node and TypeScript • Design multi-tenant systems that serve many customers from one platform, and solve the data-isolation and privacy problems that come with it properly • Build reusable agentic workflows that trigger automatically - document extraction, ingestion and workflow automation that runs without a human re-prompting it each time • Wrap the model providers into reliable production features: structured outputs, tool calls, retrieval, guardrails, approval steps and audit trails • Put in the observability that keeps an AI system honest - tracing agent runs, catching failures, and watching cost and latency • Use coding agents to move fast, and review what they produce with real engineering judgement before any of it ships • Own what breaks in production and fix the root cause, not the symptom What You'll Need Essential • Production-hardened Node and TypeScript from real commercial systems - side projects and AI-generated hobby code will not be enough • Genuine experience building multi-tenant systems that serve many customers at once, and solving the privacy and data-isolation problems that follow • A track record at the production sharp end - real scale, real users, and clear stories of what broke and what you changed because of it • Strong PostgreSQL: schema design, indexing, migrations, and fixing queries when they're slow • Around five years shipping production software, with evidence you deliver systems, not prototypes Bonus • Integrating LLM SDKs into production - very helpful, but not a blocker • Comfort with pgvector or similar for retrieval The Person You're a software engineer first with a CS background, drawn to AI because that's where the interesting problems are right now - not because you're collecting technologies. You've shipped real systems, you carry the scars, and you can tell genuine capability from hype. You work well in a small, fully distributed team where clear communication across time zones is the difference between shipping and stalling, and you take ownership without being asked. The Role • Fully remote, working UAE / Gulf hours - the team is small and spread across three continents, so overlap and clear communication genuinely matter • Around five years of production software experience • Early-stage startup: live product, paying customers, and real ownership of what you build Show more Show less

Essential: Production-hardened Node and TypeScript from real commercial systems; strong PostgreSQL (schema design, indexing, migrations); around five years shipping production software with evidence of delivering systems; genuine experience building multi-tenant systems that serve many customers with data-isolation and privacy considerations. Bonus: integrating LLM SDKs into production; familiarity with pgvector for retrieval.

Build and own the backend the whole product runs on - services, APIs, workers, queues and event-driven workflows in Node and TypeScript; design multi-tenant systems with data isolation and privacy; build reusable agentic workflows for automatic document extraction, ingestion, and workflow automation; wrap model providers into production features with structured outputs, tool calls, retrieval, guardrails, approval steps and audit trails; implement observability for reliability, including tracing, failure handling, cost and latency monitoring; use coding agents to move fast and ensure engineering judgment before shipping; own production issues and address root causes.