Grievance Portal GCP Codebase Dossier

This dossier documents the FastAPI/React grievance redressal platform and its migration from local Docker infrastructure toward GCP Cloud Run, Cloud SQL, GCS, Firebase, Secret Manager, and Terraform.

Read first

• System overview
• Agentic AI architecture
• Data model and storage
• Infrastructure and local ops
• Security

Executive summary

Citizens submit grievances through a conversational three-pass LLM workflow. The backend classifies a category, extracts mandatory fields, asks follow-ups, then creates and routes a structured case. Officers and administrators use role-specific dashboards, while a role-filtered pgvector RAG service answers policy questions. The GCP deployment provisions Cloud Run services/jobs, Cloud SQL, GCS, IAM, and secrets with Terraform.

flowchart TD
  classDef default fill:#1e293b,stroke:#38bdf8,stroke-width:2px,color:#f8fafc
  classDef highlight fill:#065f46,stroke:#34d399,stroke-width:2px,color:#f0fdf4
  Citizen["Citizen"]:::highlight --> FE["React/Vite frontend"]
  FE --> API["FastAPI API"]
  API --> DB["Cloud SQL PostgreSQL + pgvector"]
  API --> Cache["Redis session state"]
  API --> GCS["GCS attachments and ATR PDFs"]
  API --> LLM["LiteLLM / Gemini"]
  Ops["Terraform + Cloud Run jobs"] --> API
  linkStyle default stroke:#64748b,stroke-width:2px

Coverage

The chapters cover evidence, product flow, AI orchestration, backend/frontend design, storage, local/GCP operations, security, tests, engineering highlights, interview preparation, and risks.

Commands

just setup
just up-all
just migrate
just seed
just ingest-rag
just test

MkDocs preview is available with mkdocs serve -f mkdocs-dossier.yml; make consolidate creates the single Markdown report.