ApplyAI — AI Job Application Assistant

ApplyAI is a cloud-native, AI-powered assistant that helps streamline the job search. The MVP focuses on shortening application time, increasing application volume, and improving the quality of each submission by providing AI-driven resume tailoring and conversational career guidance.

Table of contents

• Overview & Goals
• Architecture & Tech Stack
• Core Features
• Data Model (Firestore)
• Getting Started (Local Development)
• Testing Suite
• Roadmap
• License

Overview & Goals

ApplyAI helps job seekers automate and improve their job applications. The MVP aims to:

• Reduce application time by automating tailored materials.
• Increase application throughput by making it easy to generate high-quality applications.
• Maximize job-fit relevance by emphasizing the most relevant skills in tailored resumes using state-of-the-art LLMs.

Architecture & Tech Stack

This project is a full-stack, decoupled application following a modern serverless-first architecture.

graph TD;
    User((User)) -->|Next.js/React| FE[Frontend Client];
    FE -->|Auth| FB_Auth[Firebase Auth];
    FE -->|JSON/REST| BE[FastAPI Backend];
    BE -->|Query/Write| FS[(Cloud Firestore)];
    BE -->|Prompt/Text| Gemini[Google Gemini AI];

Loading

To ensure VS Code recognizes the formatting immediately, I have placed the remaining content inside a single block. Copy everything from the box below and paste it directly after your Mermaid diagram.

Markdown

Stack & Rationale

• Frontend: React (Next.js 15) + TypeScript + Tailwind CSS
  • Modern, type-safe UI with efficient client-side routing and global state management via React Context.
• Auth: Firebase Authentication (Google OAuth)
  • Secure identity management that provides unique UIDs to link data across the stack.
• Backend: Python 3.12 + FastAPI
  • High-performance, asynchronous framework optimized for I/O-bound tasks like AI model inference.
• AI Model: Google Gemini 2.5 Flash
  • Utilized for both conversational "Career Coach" interactions and complex "Resume Tailoring" logic.
• Database: Google Firestore (NoSQL)
  • A serverless document database used to persist user-specific chat history and resume generation sessions.
• DevOps: GitHub Actions + Pytest
  • Automated CI/CD pipeline ensuring code quality and 100% mocked backend testing.

Core Features

1. Authentication & Security
   • Secure Google Sign-In with persistent session management.
   • Environment-driven configuration for API keys and Cloud credentials.
2. Persistent AI Chat History
   • POST /chat and GET /chats/{user_id} endpoints.
   • Career guidance that persists across sessions, allowing users to pick up where they left off.
3. Resume Tailoring Engine
   • POST /resumes and GET /resumes/{user_id} endpoints.
   • Intelligent rewriting of resumes based on job descriptions, saved to the cloud for future reference.

Data Model (Firestore)

• chats (collection)

  • doc_id (auto-generated)
    • user_id (string)
    • messages (array of { role: 'user'|'ai', content: string })
    • timestamp (serverTimestamp)

• tailored_resumes (collection)

  • doc_id (auto-generated)
    • user_id (string)
    • jobDescription (string)
    • originalResume (string)
    • tailoredResume (string - Markdown)
    • createdAt (timestamp)

Getting Started (Local Development)

Prerequisites

• Python 3.12+ / Node.js 18+
• Google Cloud Service Account with Firestore and Gemini API access.

1) Backend (server/)

1. cd server && python3 -m venv .venv && source .venv/bin/activate
2. pip install -r requirements.txt
3. Create .env with GEMINI_API_KEY and GOOGLE_APPLICATION_CREDENTIALS.
4. Run: PYTHONPATH=. uvicorn main:app --reload --port 8000

2) Frontend (client/)

1. cd client && npm install
2. Create .env.local with your Firebase config and NEXT_PUBLIC_API_URL.
3. Run: npm run dev

Testing Suite

The backend includes a robust unit testing suite using Pytest and mocker. To maintain speed and zero-cost CI, all external calls to Gemini and Firestore are fully mocked to prevent network dependency and unnecessary API costs.

To run tests:

cd server
pytest

Deployment Notes

The backend is deployed to Google Cloud Run. The frontend is currently local but configured for deployment to Vercel or Firebase Hosting.

• CORS: The FastAPI app allows requests from http://localhost:3000. For production, update server/main.py with the deployed frontend domain.

Roadmap

• Phase 1: Core AI Tailoring Logic & Form UI.
• Phase 2: Firebase Authentication & Global Context.
• Phase 3: Backend Persistence (Firestore) & Naming Convention Alignment.
• Phase 4: Mocked Test Infrastructure & History API Endpoints.
• Phase 5: Frontend Hydration (Displaying historical chats/resumes in the UI).
• Phase 6: Production Deployment (Cloud Run & Vercel).

License

See the LICENSE file for details.