Resume Modifier

Resume Modifier is a Dockerized web app for tailoring a resume and cover letter to a job posting, previewing the result, exporting polished PDFs, and keeping a per-user history of previous runs.

What the application does

• Accepts a job description, resume, and optional job URL.
• Uses a configured AI provider to generate an optimized resume, cover letter, gap analysis, interview prep, and LinkedIn outreach content.
• Stores optimization history per user account.
• Renders downloadable PDF outputs from selectable templates.
• Includes backup/restore helpers plus workflow integrations for job-scanning support.

Stack

• Frontend: static single-page app in frontend/index.html served by nginx:alpine
• Backend: FastAPI app in backend/main.py
• Database: PostgreSQL 15
• Integrations: n8n workflows and crawl4ai
• Orchestration: Docker Compose via docker-compose.yml

Services

The Compose stack starts these services:

• frontend: serves the SPA on http://127.0.0.1:${FRONTEND_PORT}
• backend: FastAPI API on http://127.0.0.1:8000
• db: PostgreSQL on 127.0.0.1:5432
• n8n: workflow automation on 127.0.0.1:5678
• crawl4ai: crawl API on 127.0.0.1:11235

All ports are bound to 127.0.0.1 only.

Repository contents

This repository intentionally keeps only the files needed to build and run the Docker stack:

Resume_modifier/
├── .env.example
├── .gitignore
├── LICENSE
├── README.md
├── docker-compose.yml
├── backend/
│   ├── Dockerfile
│   ├── main.py
│   └── requirements.txt
├── frontend/
│   ├── admin_icon.svg
│   ├── favicon.svg
│   └── index.html
└── n8n/
    ├── gen_v5_workflows.py
    └── workflows/

Local-only folders such as .vscode, .env, Raw_Files, generated backups, and manual migrations are excluded from version control.

Quick start

Prerequisites

• Docker Desktop or Docker Engine with the Compose plugin
• An AI provider endpoint reachable by the backend container, such as LM Studio or Ollama

Setup

1. Clone the repository.
2. Copy .env.example to .env.
3. Adjust FRONTEND_PORT if needed.
4. Start the stack:

docker compose up --build -d

5. Open the frontend at http://localhost:<FRONTEND_PORT>.

To stop the stack:

docker compose down

To remove volumes as well:

docker compose down -v

Environment variables

Use a local .env file for overrides. The committed example file documents the supported values.

Variable	Purpose
TIMEZONE	Time zone used by the containers
FRONTEND_PORT	Host port for the frontend container
N8N_WEBHOOK_SECRET	Shared secret for backend to n8n webhook requests
N8N_ENCRYPTION_KEY	n8n encryption key
CRAWL4AI_TOKEN	API token for the crawl4ai service

Note: on some Windows machines, port 4310 can fall inside a reserved range. If that happens, use another local port such as 3210 in your .env file.

Application flow

1. Register or sign in.
2. Open Settings and choose lmstudio or ollama.
3. Configure the provider URL, model, and optional API key.
4. Run an optimization from the main form.
5. Review results in the app, then export resume or cover-letter PDFs.
6. Revisit prior runs from History.

Notes

• The backend stores user accounts, history, and encrypted provider API keys in PostgreSQL.
• The app seeds a default admin user on first run if the database is empty.
• Manual SQL files under migrations/ are not required to build or run the Docker stack, so they are left out of version control for this repo.

LM Studio

1. Download and launch LM Studio
2. Load a model and start the local server (default: http://localhost:1234)
3. In the app Settings, set:
   • Provider: lmstudio
   • Base URL: http://host.docker.internal:1234 (the app will auto-strip a trailing /v1 if you include it)
   • Model: copy the model ID shown in LM Studio

Note: The application prefers LM Studio's native REST API endpoints (/api/v1/chat, /api/v1/models) when provider is set to lmstudio. If those endpoints are not available, the backend will fall back to LM Studio's OpenAI-compatible endpoints under /v1/* (e.g., /v1/chat/completions). You can safely point the Base URL to either the host root (e.g. http://host.docker.internal:1234) or to the OpenAI-compatible path (e.g. http://host.docker.internal:1234/v1) — the backend will detect and use the best supported API shape.

If your LM Studio server is configured to require an API token, enter that token in the app Settings API Key field (the app sends it as an Authorization: Bearer <token> header). The /models/detect and other calls will return 401 Unauthorized unless a valid token is provided.

Ollama

1. Install and start Ollama: ollama serve
2. Pull a model: ollama pull llama3
3. In the app Settings, set:
   • Provider: ollama
   • Base URL: http://host.docker.internal:11434
   • Model: llama3 (or whichever model you pulled)

OpenAI / Other Compatible APIs

Set the provider to lmstudio (uses the OpenAI-compatible client path), point the Base URL at the endpoint, and supply a real API key.

---

Usage

1. Register / Login

On first launch, click Register and create an account. All data is stored locally in the PostgreSQL container. Log in with your credentials to access the optimizer.

2. Configure AI Settings

Go to the Settings tab and fill in your AI provider details (see above). Click Save Settings. You can also click Detect Models to auto-fetch available models from your inference server.

3. Optimize Your Resume

1. Go to the Resume Optimizer tab.
2. Paste the Job Description into the left text area.
3. Either:
   • Paste your current resume as Markdown in the right text area, or
   • Upload a PDF (it will be extracted automatically).
4. Click Optimize Resume.
5. A "DO NOT REFRESH THIS PAGE" warning appears in the header while optimization is running.
6. When complete, a browser notification is sent and the results are displayed.

The output persists in the current browser session — refreshing the page will restore your last optimization result. Results are cleared only by clicking Clear.

4. Preview and Download PDF

1. After optimization, click Download PDF (or the template thumbnail in history).
2. A modal opens with a live preview on the right and action buttons (template selector + download) at the top.
3. Select a resume template from the dropdown.
4. Preview updates automatically.
5. Click Download PDF to save. The modal stays open so you can switch templates and download again.

5. Generate a Cover Letter

After optimizing your resume, click Generate Cover Letter. The AI uses the same job description and resume to produce a tailored cover letter. Download it as a PDF using any of the cover letter templates.

6. Optimized Resumes History

The Optimized Resumes tab shows all past optimization runs. You can:

• Search by job title or company
• Filter by date range
• Click any entry to view the full optimized resume
• Choose a per-entry template and download its PDF
• Delete individual entries

7. Backup and Restore

In the Settings tab:

• Create Snapshot — saves a compressed backup of the database to /app/data/backups/ inside the container (mapped to a named Docker volume).
• Restore Snapshot — lists available snapshots; click one to restore the database to that point in time.

---

Resume Templates

Template	Style
Classic	Clean, corporate; uppercase section headers; dark navy name
Modern	Garamond serif font; reduced letter-spacing; elegant proportions
Executive	Bold orange accent on section dividers; authoritative layout
Minimal	Uppercase slate-gray section labels; maximum whitespace
Tech	Blue left-border accent on name; monospace-adjacent feel
Creative	Blue left-bar on section headers; expressive but professional

All templates share:

• Name rendered at 27pt, left-aligned
• Contact info (Phone → Email → Address → LinkedIn) at 11pt, right-aligned
• Consistent page margins and line-height for broad compatibility

---

Cover Letter Templates

Template	Style
Classic	Formal business letter format
Modern	Contemporary with tasteful typography
Executive	Premium look matching the Executive resume template
Minimal	Ultra-clean, distraction-free

---

PDF Export

PDFs are generated server-side using WeasyPrint 62.3 (with pydyf 0.11.0 pinned for compatibility).

The backend parses the Markdown resume header to extract structured contact fields:

• Name — first non-empty line
• Phone — line matching a phone number pattern
• Email — line containing @
• Address — line containing a comma or common location keywords
• LinkedIn — line containing linkedin

These are rendered into a two-column <header> block: name on the left, contact details stacked on the right. The rest of the resume body follows below.

Template-specific CSS overrides (fonts, colors, accent styles) are injected per-request.

---

API Reference

All endpoints are on the backend at http://localhost:8000.

Method	Path	Description
POST	/register	Create a new user account
POST	/login	Authenticate; returns JWT token
POST	/change-password	Change password for authenticated user
POST	/optimize	Run AI resume optimization
POST	/generate-cover-letter	Generate a cover letter
POST	/pdf/render	Render Markdown to PDF (returns binary)
GET	/history	List all optimization history entries
POST	/history/{document_id}/template	Set the template for a history entry
DELETE	/history/{document_id}	Delete a history entry
POST	/models/detect	Detect available models from the AI server
POST	/backup/snapshot	Create a database backup snapshot
GET	/backup/snapshot	List available backup snapshots
POST	/backup/restore	Restore the database from a snapshot
POST	/provider-api-key	Store or update an encrypted provider API key for the current user (form: provider, api_key)
DELETE	/provider-api-key	Remove a stored provider API key (query: provider)
GET	/provider-api-keys/status	List supported providers and whether a key is stored for the current user
POST	/resume/store	Upload a resume PDF; server extracts and stores plain text (multipart/form-data)
GET	/resume/stored	Return stored resume preview and parsed profile for the current user
POST	/resume/profile/ai	Run an AI model against the stored resume to infer a structured profile (form params: lm_url, lm_model, provider, api_key)
GET	/provider-api-key	Retrieve the decrypted provider API key for the current user (query: provider)
PATCH	/history/{document_id}/resume	Update the tailored_resume content for a history entry (form: tailored_resume)
PATCH	/history/{document_id}/cover-letter	Update the cover_letter content for a history entry (form: cover_letter)
DELETE	/resume/stored	Delete the stored resume for the current user

Interactive API docs (Swagger UI) are available at http://localhost:8000/docs while the backend is running.

---

Ports

Service	Container Port	Host Binding
Frontend (nginx)	80	127.0.0.1:4310 (configurable via FRONTEND_PORT)
Backend (FastAPI)	8000	127.0.0.1:8000
Database (PostgreSQL)	5432	127.0.0.1:5432

All ports are bound to 127.0.0.1 — accessible only from the local machine.

---

Development Notes

• The frontend is a single HTML file (frontend/index.html) served as a static file by nginx. No build step is required — edit and refresh.
• The backend mounts ./backend as a volume, so Python file changes take effect after docker compose restart backend.
• weasyprint==62.3 and pydyf==0.11.0 are pinned together. Upgrading either independently may break PDF rendering.
• The openai Python package is used as a generic HTTP client for all AI providers (not just OpenAI) via the base_url parameter. This means any OpenAI API-compatible service works out of the box.
• Tab state, optimization results, and AI settings are persisted in sessionStorage (tab + results) and localStorage (AI settings) so they survive page refreshes within the same session.

---

Security Notes

• All ports are bound to localhost only — not exposed to external networks.
• Passwords are hashed with bcrypt via passlib.
• JWT tokens are used for all authenticated API calls.
• The default database password in docker-compose.yml (securepassword123) should be changed before any shared or production deployment.
• No telemetry or external calls are made — all AI requests go to your locally configured inference server.