Traditional employee onboarding wastes up to 40% of an experienced hire's time by forcing them through redundant compliance and training modules they already know, while simultaneously overwhelming freshers with gaps in their foundational knowledge. The "one-size-fits-all" learning approach is fundamentally broken for modern agile teams.

SkillForge is an AI-driven adaptive onboarding engine that personalizes training pathways from day one. By instantly cross-referencing a candidate's Resume against a target Job Description (JD), our ML pipeline surgically extracts exact skill gaps and instantly generates a dynamic, gamified learning roadmap tailored uniquely to their career delta.

• 🧠 Interactive Skill Gap Map: A dynamic react-flow topology mapping your Missing vs Required proficiency levels, styled with color-coded severity boxes (Critical, Partial, Near-Competent).
• 🗺️ Gamified "Candy Crush" Roadmap: A highly interactive, S-curve learning pathway visualizing your journey from baseline to target role.
• 📄 Native 6-Page Document Generation Engine: A robust exporter powered by jsPDF that builds a custom AI-curated syllabus with YouTube links, educator recommendations, and weekly study plans.
• 🎥 Curated 6-Video YouTube Arrays: Dynamically fetching and sorting the top global educator channels (CodeWithHarry, freeCodeCamp) directly into interactive grid tiles.
• 🛡️ Gemini API Token Load Balancer: A dual-pronged token consumption engine that surgically compresses resume payloads by 80% and seamlessly cascades to gemini-1.5-flash to effortlessly bypass strict Google Rate Limits.
• ⚔️ Asynchronous Multiplayer Quiz Arena: Challenge peers in real-time with host-controlled waiting rooms, ending with a detailed post-match question-by-question predictive analytics dashboard.
• 📓 Condensed NotebookLM-Style Mentor: An integrated AI sliding pane that ingests video URLs and spits out brief, highly-actionable learning summaries & quizzes instantly.

graph TD
    subgraph Frontend [React Multi-View SPA]
        A[Authentication: Clerk] --> B[Upload UI]
        B --> C[Skill Gap Map / React Flow]
        C --> D[LMS Dashboard Hub]
        D --> E[PDF Export Engine]
    end

    subgraph Backend [FastAPI Server]
        B -.-> |Multiform Resume & JD| F(FastAPI Endpoints)
        F --> G{Gemini 2.5 Flash API}
        F --> H[TensorFlow Emdeddings]
        G -.-> |Vectorize & Map| I[Gap Analysis Engine]
    end

    subgraph Data & Storage
        I -.-> J[(Supabase DB: User Metrics)]
        I -.-> K[YouTube v3 API]
    end

    I --> |Returns Topological JSON| C

sequenceDiagram
    actor You
    participant UI as React Frontend
    participant Server as FastAPI Backend
    participant AI as Gemini 2.5 NLP
    participant YT as YouTube API

    You->>UI: Uploads Resume (PDF) & Job Title
    UI->>Server: POST /api/analyze-gap
    Server->>AI: Prompts text extraction & JD mapping
    AI-->>Server: JSON: Critical, Partial, Minor Gaps
    Server->>YT: Query canonical video resources per gap
    YT-->>Server: Video IDs & Metadata
    Server-->>UI: Returns comprehensive Roadmap tree
    UI->>You: Renders Interactive Skill Gap Map
    You->>UI: Clicks "Generate Plan"
    UI->>You: Warps to LMS Dashboard + YouTube Auto-Play

Copy the .env.example file to frontend/.env and backend/.env.

cp .env.example frontend/.env

cd backend
python -m venv venv
venv\Scripts\activate   # Windows
pip install -r requirements.txt
uvicorn main:app --reload --port 8000

cd frontend
npm install
npm run dev