CVMorph
Enterprise AI-Powered CV Transformation & Branded Resume Formatting Platform.
Ingest candidate resumes in any format (PDF, Word, or raw text) → Extract into a canonical, confidence-scored profile using AI → Polish in Review Studio with conversational AI Copilot & job tailoring → Render into branded, pixel-perfect .docx and .pdf resumes.
Key Features
- Multi-Source Intake: Upload raw PDFs and DOCX files or paste unstructured text backgrounds with target role positioning.
- AI Extraction & Provenance Tracking: Extracts structured career summaries, grouped technical competencies, chronologically formatted employment, and education using Google Gemini 3.5 Flash. Every field carries confidence scores and traceable evidence spans.
- Strategic Role Alignment & Bluffing Engine: 4-tier granular alignment slider (
Strict Facts, Role Focus, Role Alignment, Scope Expansion) to adapt candidate experience directly against a target Job Description with audit trails for extrapolated claims.
- Interactive Review Studio: In-browser side-by-side editing, low-confidence flag verification, and instant inline candidate updates.
- AI Profile Copilot: Built-in conversational sidecar enabling recruiters to command complex transformations (e.g. "Condense summary to 3 high-impact bullets", "Quantify bullet points with latency and throughput metrics", "Add Docker and Kubernetes to DevOps").
- Multi-Profile Branching & Master Profiles: Generate multiple tailored variations of a candidate's CV for different roles, elect a master default profile, and approve versions independently.
- Deterministic Word & PDF Generation: Generates clean
.docx files using Jinja-templated XML document manipulation, and converts to .pdf via headless LibreOffice.
- Zero-Egress Object Storage: Cloudflare R2 integration for ultra-fast, cost-effective document storage and presigned download URLs.
System Architecture & Tech Stack
cvmorph/
├── apps/
│ ├── api/ # FastAPI Backend (Python 3.12, SQLAlchemy, Alembic, LibreOffice)
│ └── web/ # Next.js 16 Web Application (React 19, TypeScript, Tailwind CSS)
├── packages/
│ └── shared-types/ # Shared TypeScript types derived from canonical JSON schema
├── docs/ # Product specifications, API reference, schema mapping contracts
├── infra/ # PostgreSQL initialization scripts
├── Dockerfile # Authoritative API container Dockerfile
├── vercel.json # Monorepo deployment and routing configuration
└── DEPLOYMENT.md # Production deployment and infrastructure guide
| Layer | Technologies |
|---|
| Frontend | Next.js 16, React 19, TypeScript, Tailwind CSS, TanStack Query, Lucide Icons, Sonner |
| Backend | FastAPI, Python 3.12, Pydantic v2, SQLAlchemy (asyncpg), Alembic, docxtpl |
| Document Processing | PyMuPDF, headless LibreOffice, python-docx, poppler-utils |
| AI / LLM | Google Gemini 3.5 Flash (google-genai), Claude Structured Outputs compatible |
| Database | PostgreSQL 16+ (Supabase / local Docker) with multi-tenant row-level security |
| Storage | Cloudflare R2 (S3-compatible) with zero egress bandwidth fees |
| Package Management | Python: uv · Node.js: npm |
Local Development Quickstart
Prerequisites
1. Clone the Repository
git clone https://github.com/your-org/cvmorph.git
cd cvmorph
2. Environment Configuration
Copy sample environment files:
# Backend environment
cp apps/api/.env.example apps/api/.env
# Frontend environment
cp apps/web/.env.local.example apps/web/.env.local
Fill in your GEMINI_API_KEY, database credentials, and Cloudflare R2 credentials in apps/api/.env.
3. Run with Docker Compose (Recommended)
Launch the entire backend stack with PostgreSQL, automated migrations, system template seeding, and LibreOffice conversion:
docker compose up --build
The API will be accessible at http://localhost:8000. Auto-generated OpenAPI documentation is available at http://localhost:8000/docs.
4. Run the Frontend
In a separate terminal:
cd apps/web
npm install
npm run dev
Open http://localhost:3000 in your browser.
Production Deployment
Both the Next.js frontend and containerized FastAPI backend deploy together onto Vercel with a single command:
vercel --prod
The monorepo uses vercel.json to route both services under the exact same domain, eliminating CORS preflights and cookie restrictions.
Refer to DEPLOYMENT.md for the complete cloud setup guide (Supabase PostgreSQL, Cloudflare R2, Google Gemini 3.5 Flash, and Vercel).
Documentation Index
License
Internal proprietary software. All rights reserved.
Ashish Kumar
Shubham Dubey
