A high-level architectural walkthrough.
End-to-End Infrastructure Flow
Cloudflare Edge Proxy
Google Cloud Run (v2)
SQLite + Litestream & GCS
Google AI Studio
Component Architecture Breakdown
An air-gapped on-premise document automation platform built for corporate legal, HR, and finance departments. Merges datasets into pre-approved Word templates locally with zero data egress.
A high-throughput multimodal document parsing engine converting unstructured invoices and scanned forms into strictly typed Pydantic models with dynamic runtime schemas.
An in-memory resume building and PDF compilation platform with bidirectional visual canvas editing, live HTML/CSS synchronization, and rich layout parsing.
Infrastructure-as-Code monorepo providing reproducible staging and production environments on GCP with zero long-lived service account credentials.
Complete Technology Matrix
| Domain / Layer | Technologies & Libraries | Architectural Purpose |
|---|---|---|
| Cloud & Serverless | Google Cloud Run (v2), Artifact Registry, Cloud Functions | Container execution, auto-scaling, scale-to-zero when idle |
| Edge & DNS | Cloudflare WAF, Transform Rules, Zero Trust OTP | Edge rate limiting, SSL termination, ingress secret verification |
| Backend & API | Python 3.13, FastAPI, Django 5, Uvicorn, Gunicorn | REST endpoints, HTMX HTML partials, session and OTP management |
| AI & Processing | Gemini 2.5 Flash, google-genai, WeasyPrint, docxtpl | Multimodal document parsing, in-memory PDF/Word compilation |
| Data & Storage | SQLite, Litestream, Google Cloud Storage, Firestore | Continuous SQLite replication, temporary upload buckets, licensing DB |
| DevOps & Security | Terraform, GitHub Actions, Workload Identity (OIDC), Secret Manager | Keyless CI/CD deployments, remote state locking, secret injection |