AI Integration + Full-Stack Engineer | Agents, APIs, Polished Front End (LATAM, Ongoing)

We're a small digital growth agency (paid media, SEO, creative) working with DTC, ecommerce and local brands. A lot of our operations already run on AI: assistants connected to business tools through MCP, a Python data platform in the cloud, and internal dashboards. We want one sharp engineer who can own this end to end and make it fast, reliable and pleasant to use. Who you are: - A real full-stack engineer: you design the backend, the data model and the API, and you also care how the front end looks and feels. Clean UI and good UX matter to you. - Deeply into AI. You try new models and tools the week they ship, you have opinions on which frontier model is best for which job, and you build with agents, MCP and coding assistants every day. - You follow the AI builder conversation on X and know what shipped this week and what actually works. - Polished and clear. Your written English is strong, your updates are short and specific, and you finish what you start. What you'll work on: - Build and improve AI agent workflows and MCP connectors for common business tools: team chat, email and docs, project boards, CRM, accounting and ad platforms. - Own Python data pipelines and scheduled cloud jobs: ingestion, validation, reporting. - Ship internal tools and small web apps (React/Next.js, TypeScript) the team actually enjoys using. - Make automations dependable: approvals before sensitive actions, retries, no duplicates, failure alerts, proof a job really finished. - Write short docs so the team can use and maintain what you build. Must have: - Python and TypeScript; React/Next.js; SQL; REST/GraphQL APIs, OAuth and webhooks. - At least one LLM or agent feature you personally shipped that real people use. - Good judgment on permissions, credentials and human review for client-facing or financial actions. - Individual freelancer, not an agency. You personally do all the work. Nice to have: MCP servers, Google Cloud (Cloud Run, BigQuery), ad platform APIs (Google Ads, Meta), a public GitHub or X presence where you share what you build. How we start: a small paid trial with scope and hours agreed upfront, then ongoing. Communication, contracts and payments stay on Upwork. To apply: answer the screening questions, link 2 things you built (repo, demo or Loom) and say what you personally did on each.