Description
Summary We are looking for an experienced AI Engineer / Full-Stack Automation Developer to build a custom AI-powered marketing agent. The agent will streamline and partially automate our ad operations across Google Ads and Meta Ads (Facebook & Instagram). The ideal candidate has hands-on experience integrating LLMs with marketing APIs, setting up automated workflows, and building safe, human-in-the-loop advertising systems. Key Responsibilities: API Integration: Connect LLM frameworks with the Meta Marketing API and Google Ads API for reading analytics and drafting campaigns. Ad Copy & Creative Generation: Build workflows using OpenAI GPT-4o / Claude 3.5 to auto-generate platform-compliant ad headlines, primary text, and ad creative prompts based on brand guidelines. Performance Monitoring Agent: Create automated scripts to fetch daily performance metrics (CTR, CPC, CPA, ROAS) and generate summary insights or slack/email alerts. Draft & Budget Automation: Programmatically generate campaign structures (ad sets, keyword targeting, copy) and publish them as Paused Drafts for human approval. Guardrails & Safety Rules: Implement strict budget caps, policy compliance checks, and approval triggers to prevent unauthorized spending or ad rejections. Required Skills & Qualifications: API Expertise: Proven experience working with Google Ads API and Meta Marketing API (OAuth 2.0, developer tokens, campaign/ad group creation, reporting endpoints). AI/LLM Development: Experience with OpenAI API, Anthropic Claude API, LangChain, LlamaIndex, or custom prompt engineering pipelines. Automation Frameworks: Fluency with Python, Node.js, or low-code orchestration platforms like n8n or Make.com. Database & Storage: Basic setup in Supabase, PostgreSQL, or vector databases (Pinecone) to store historical ad performance and brand assets. Project Phases & Scope: Phase 1 (MVP): Read-only integration to pull ad performance data, generate AI insights, and draft ad copy variations. Phase 2: Automated draft campaign creation directly inside Google Ads Manager and Meta Ads Manager (human review required before publishing). Phase 3: Automated performance rules (e.g., auto-alerting or auto-pausing underperforming ads based on strict KPIs).