AI Integration and Automation Engineer (APIs, n8n, Claude Code, Data Pipelines)

AI Integration and Automation Engineer (APIs, n8n, Claude Code, Data Pipelines) Stack Growth Solutions builds modern finance back offices for private equity backed companies. We are hiring an engineer to design and build the integration and AI layer across our clients: data pipelines between accounting, payroll, operations, and reporting systems, plus AI agents and automations that run real business processes. You will work directly with the firm's owner. This is not an easy role and the pay rate reflects that. We want someone who has built serious, production-grade automation in demanding environments. Our clients are companies being upgraded from the ground up. The systems you build replace spreadsheets, email chains, and manual work, and they have to run reliably without you watching them. You will work closely with the firm's owner and client leadership on system design, business workflow, and operations, but we are hiring someone with an entrepreneurial mindset: you study how a business actually runs, map how data should flow through it, see where the process is broken, and design the system that fixes it. You will not be handed a complete spec every time. You must also be a strong communicator, as you will be discussing designs and workflows with the owner and clients directly. You will own: - Designing how data flows across each client's stack: accounting, payroll, field operations, CRM, reporting, and project tools, regardless of which specific software sits at each node - Building and maintaining integrations via APIs, webhooks, middleware (n8n, Make, Zapier), and custom code where connectors do not exist - Advanced AI implementation: Claude Code, agents, structured LLM workflows, and tool-connected automations embedded in accounting and back office processes, built with validation and human oversight so they are dependable - ETL-style work: extracting, cleaning, transforming, and syncing data between systems so reporting is built on reliable data - Workspace and board buildouts in tools like Monday.com and Notion; we consider this baseline work you should find trivial – should utilize Claude connectors to streamline this work - Hardening one-off builds into reusable templates and playbooks we deploy across clients The bar: - 3+ years building automations and integrations in production, with n8n, Make, Zapier, or code, and role where you had accountability as these broke - Strong API engineering: REST, webhooks, OAuth and key auth, JSON, pagination, rate limits, retries, and error handling; you read documentation without being walked through it - Real data engineering instincts: you think in sources, transformations, and destinations, and you can explain how you keep two systems in sync without silent drift - Advanced hands-on LLM work: Claude Code, agents, structured outputs, or tool-connected workflows in production; a chatbot demo does not qualify - Scripting ability (Python or JavaScript) for the gaps middleware cannot cover - You can map a messy real-world process nobody documented, ask sharp questions early, and turn it into a system without supervision - Self-motivated and comfortable with ambiguity; you propose the design rather than wait for one - At least 6 hours of daily overlap with US Eastern time Finance or accounting context is a plus, not a requirement. We supply the accounting judgment; you supply the architecture and flawless execution. Process: application, live portfolio walkthrough, paid test project. We hand you a spec for a system we have already built and see how close you get in a week. Begin your application with the word PIPELINE. To apply, answer the following in your cover letter. Answers without specifics will not be considered. - Describe the most complex integration or automation system you have built end to end: the systems involved, how data flowed between them, what broke most often, and what you would rebuild differently today. - Describe an integration you built where no native connector existed: which APIs, which auth method, how you handled failures, rate limits, and retries, and how you knew when it silently failed. - Describe the most advanced LLM-based system you have built that was not a chatbot: what it did, how you made it reliable enough for someone to depend on daily, and what validation or oversight you built around it. - Two systems both claim to be the source of truth for the same records and they disagree. Walk through how you would design the sync and reconciliation so drift gets caught and escalated. - Your hourly rate, weekly availability, and working hours in US Eastern time.