Summary I am building an AI-powered system to automate geotechnical engineering reports for a civil engineering firm. The system ingests field/lab data (PDF borehole logs, granulometry tests), runs engineering calculations, and generates 50–150 page technical reports using LLMs and a historical-report knowledge base. Current state Existing codebase (FastAPI backend, React/TypeScript frontend, PostgreSQL, Alembic) with CRUD, auth, data import, and basic charts already built. You will extend it with the AI and report-generation core. What you'll build LLM-driven narrative generation (Claude/OpenAI) with RAG over historical reports (pgvector) PDF parsing pipeline for structured geotechnical templates (Camelot/Tabula + LLM vision fallback) Geotechnical calculation engine (bearing capacity, settlements, etc. — formulas provided by the in-house engineer) Word/PDF report generator matching an existing 100+ page template (python-docx + matplotlib charts) Editor UI for engineer review/approval before export Required 3+ years Python + FastAPI 3+ years React + TypeScript Production experience with LLM APIs (OpenAI or Anthropic) + RAG Demonstrated experience generating complex Word/PDF documents programmatically PostgreSQL + SQLAlchemy + Alembic Bonus Civil engineering, AEC, or scientific document automation background PDF table extraction (Camelot, Tabula, or LLM-based) Budget $300~500