AI Automation Engineer — AIOS Agent & Automations

AI Automation Engineer — AIOS Agent Workflows & Automations About Us We build AI-powered SaaS products and internal operating systems. Our flagship platform, AIOS (AI Operating System), runs persistent AI agents across all standard departments of a tech and service based business types. We operate at high deployment velocity and need additional AI Automation Engineers to own execution of PRD-driven agent features as we scale. The Role You'll build and ship AI agent workflows and automations from PRDs I provide. You work directly with me as the product owner. This is a long-term hourly engagement for the right person. This is not a prototype or demo role. Every feature you build runs in production on a live system with real users. We need someone who reads a spec, asks the right questions before touching code, and ships clean on the first submission. The Stack - Python (primary language — FastAPI required) - DigitalOcean VPS (Ubuntu 24.04) - Claude Code + Hermes Agent - Neon PostgreSQL — multi-tenant, schema-per-tenant isolation - Qdrant (vector search) - Gmail, Google Calendar, Telegram, Todoist, Obsidian, Telynx, Voice Agents, Fireflies, etc. - GitHub Actions (CI/CD) - GitHub for version control - Slack for Daily communication What You'll Own - Building persistent, always-on AI agent workflows on Linux VPS - Implementing new automations and integrations from PRDs - Multi-tenant architecture — isolated data per tenant, RBAC, secure onboarding - Dashboard features for non-technical users to manage integrations and agent activity - OAuth and API key onboarding flows per tenant - Daily Slack check-ins with documented progress - PR-based workflow — branch → PR → review → merge. Nothing ships without approval. What We're Looking For - 3+ years Python development with production AI agent experience - Must have 200+ hours logged on Upwork - Must have 98% or higher Job Success Score - Must have 5+ completed jobs with ratings - Must have an average response time of 8 hours or less - Production AI agents on Linux VPS — not demos, not n8n flows - Neon Postgres or equivalent serverless Postgres with multi-tenant isolation - Dashboard development for non-technical users - Claude Code experience — CLAUDE.md, context management, disciplined builds -Contractors are never provided live credentials. All development uses environment variable stubs and synthetic data. You must be comfortable building and testing against stubs until credentials are provisioned through the proper channel - Available Monday through Friday with consistent, reliable output. Must be available for weekly online meeting during US time zone working hours. - No vibe coding — every line reviewed and understood before committing - Long-term commitment — we are not looking for a one-off - Must speak english and be able to be understood. Must read english Time Tracking Accurate time tracking via Upwork's time tracker is required and is a condition of payment. Screen captures must reflect active, project-relevant work. Manual time entry is not accepted under any circumstances. To Apply All five required or your proposal will not be reviewed: 1. Describe one specific production AI agent you built on a Linux VPS. Name the stack, describe the architecture, and explain what broke and how you fixed it. 2. Describe your experience with multi-tenant SaaS architecture. How did you handle tenant data isolation at the database level? 3. Have you built admin dashboards or onboarding flows for non-technical users? Describe what you built and how you handled OAuth or API key management securely. 4. Walk me through how you use Claude Code on a real project — CLAUDE.md structure, context management, and how you handle it when a build goes off-track. If you prefer Cursor or another tool, say so honestly. 5. Share a link to your GitHub profile. We will review your repositories and commit history as part of the evaluation.