AI Engineer – Building AI Tools for TableTap (Python/Django)

TableTap is a restaurant management platform (Django + Next.js). We're building out a set of AI agents that live in apps/ai/ and want to start with the Menu Engineering Assistant a read-only feature with no new tables and no human-in-the-loop UI, so it's a clean first build. What it does: Classifies every menu item into the classic 2×2 matrix (stars, plowhorses, puzzles, dogs) based on margin × volume Generates a weekly report emailed to the owner with 3 concrete, specific actions (e.g., "promote the Tibs, retire the chicken sandwich, raise Macchiato price by 5 Birr") What you'll build: An agent orchestrator under apps/ai/agents/ following our existing pattern: LLM + a defined tool set + system prompt + goal Tools (if not already present) under apps/ai/tools/ as DRF views with JSON-schema definitions query_top_items and query_historical_traffic A weekly scheduled job that runs the agent and sends the report via email Integration with our existing apps/ai/providers/base.py abstraction so the model can be swapped later. Audit logging via our existing audit_logs infra every agent run logs model, tokens, cost, tool calls Requirements: Strong Python + Django (DRF) Experience with LLM tool-calling / function-calling (Gemini API a plus) Comfortable designing precise JSON-schema tool definitions sloppy descriptions cause the LLM to call the wrong tool Can work from a written architecture doc and follow existing conventions rather than inventing new ones.