Career Atlas

Career Atlas (formerly Smart Resume AI) is a production-grade Candidate Optimization Platform & Agentic Research Assistant.
It combines a RAG (Retrieval-Augmented Generation) pipeline for deep internal analysis with autonomous web agents to provide real-time job market intelligence.

Status: 🚧 Active Building Phase / Beta

Powered by Google Gemini 2.0 Flash, ChromaDB, and Live Web Search.

---

🚀 Features

💎 Quick Essentials (Strategy Tools)

Foundational tools to polish your professional profile using internal benchmarks and logic.

• Job Fit Analysis: Matches your resume against market standards and highlights gaps using RAG against a database of 2400+ resumes.
• Executive Summary: Generates a punchy, professional bio for your profile.
• Resume Enhancer: Tailors your resume keywords for a specific target role (e.g., "Product Manager").
• Cover Letter: Drafts a personalized cover letter strategy for a specific job application.

⚡ Agentic Deep Research (Live Web)

Autonomous agents that go beyond static data to fetch real-time market insights.

• Job Scout: Scours the live web using targeted ATS algorithms (Greenhouse, Lever, etc.), filtering out spam/aggregators to find verified, direct-apply opportunities.
• Agentic Interview Prep: Researches specific company engineering blogs and fresh 2024/2025 community discussions (Reddit/Blind) to predict technical questions.

🧠 Core Intelligence

• DeepResearchAgent Architecture: Dedicated Python agents that autonomously plan, search, and verify web data.
• Context Command Center: Persistently holds your resume context to inform every tool.
• Multi-Channel Chat Interface: Modular Floating Assistant with distinct tool sessions.
• Image / OCR Support: Seamlessly extracts text from .jpg, .jpeg, and .png resume images using Gemini Vision.

---

🛠 Tech Stack

Frontend

• React 18 (Vite)
• Context API for Global State & Chat History
• Lucide React for Modern Iconography
• CSS Modules: Custom "Glassmorphism" Design System
  • Theme 1: Shining Cyber Blue (Essentials)
  • Theme 2: Emerald Power (Agentic Tools)

Backend

• FastAPI (Python 3.10+)
• ChromaDB (Vector Database - for Resume Benchmarking)
• Google Gemini API (LLM - Gemini 2.0 Flash)
• DuckDuckGo Search (Web Tools for Agentic Workflows)
• PDFMiner (Text Extraction)

---

📂 Project Structure

career-atlas/
├── app/                        # Backend (FastAPI)
│   ├── api/                    # API Routes tailored for each agent
│   │   ├── routes/             # Endpoints (job_fit, interview, web_agents, etc.)
│   ├── core/                   # Config & Settings
│   ├── generation/             # Gemini LLM Clients & Agent Instructions
│   ├── ingestion/              # Data Processing Pipeline
│   ├── services/               # Agentic Services (job_scout, market_research)
│   └── main.py                 # App Entrypoint
├── data/                       # Dataset Storage
├── frontend/                   # Frontend (React)
│   ├── src/
│   │   ├── components/         # Chat Interfaces & UI Cards
│   │   ├── context/            # ChatContext (Message Routing)
│   │   ├── pages/              # Workbench (Main Dashboard)
│   │   └── services/           # API Client
└── scripts/                    # Utility Scripts

---

🏗 Architecture

RAG & Agentic Pipeline

The system uses a dual-engine approach: Internal Knowledge (RAG) for benchmarking and External Knowledge (Agents) for market data.

graph TD
    User[User] -->|Uploads Resume| Frontend
    Frontend -->|POST /api/ingest| Backend[FastAPI Backend]
    
    subgraph "Internal RAG Engine"
        Backend -->|Extract Text| PDFMiner
        PDFMiner -->|Generate Embedding| Gemini[Gemini 2.0 Flash]
        Gemini -->|Vector| ChromaDB[(ChromaDB Vector Store)]
        ChromaDB -->|Retrieve Top K Matches| Context[Context Window]
    end
    
    subgraph "Agentic Web Engine"
        Backend -->|Trigger Agent| Agent[Job/Interview Agent]
        Agent -->|Search Query| DDG[DuckDuckGo Search]
        DDG -->|Live Results| Context
    end
    
    Context -->|Augmented Prompt| Gemini
    Gemini -->|Streaming Response| Frontend

Loading

---

⚡ Setup & Installation

1. Prerequisites

• Python 3.10+
• Node.js 18+
• Google Gemini API Key
• ChromaDB Credentials (Optional, defaults to local if not configured)

2. Environment Variables

Create a .env file in the root directory:

GEMINI_API_KEY=your_key_here
LOG_LEVEL=INFO

# Optional: Cloud Vector DB
CHROMA_API_KEY=your_chroma_key
CHROMA_TENANT=default_tenant
CHROMA_DATABASE=default_database

3. Backend Setup

# Install dependencies
pip install -r requirements.txt

# (Optional) Ingest Seed Data for RAG
python scripts/ingest_data.py

4. Frontend Setup

cd frontend
npm install

---

🏃‍♂️ Running the Application

Start Backend

uvicorn app.main:app --reload --port 8000

API Docs available at: http://localhost:8000/docs

Start Frontend

cd frontend
npm run dev

App running at: http://localhost:5173

---

🔗 API Reference

Method	Endpoint	Description
POST	/api/parse/resume	Ingest PDF/Text resume
POST	/api/job-fit	RAG-based Fit Analysis
POST	/api/interview/agentic	Live Web Interview Research
POST	/api/find-jobs	Live Web Job Search Agent
POST	/api/enhance	Resume Enhancement
POST	/api/cover-letter	Cover Letter Draft
POST	/api/ai-assistant/chat	Career Atlas General Assistant