flowchart TB
    subgraph Client["Client Layer"]
        Browser[Browser]
        Mobile[Mobile]
    end

    subgraph Vercel["Vercel Edge"]
        NextJS[Next.js 16<br/>App Router]
        API[API Routes]
        Edge[Edge Functions]
    end

    subgraph Services["External Services"]
        Gemini[Google Gemini<br/>AI Chat]
        LinkedIn[LinkedIn API<br/>Share Posts]
    end

    subgraph GitHub["GitHub Infrastructure"]
        Actions[GitHub Actions<br/>CRON Scheduler]
        Repo[(Repository<br/>Data Warehouse)]
    end

    subgraph ML["ML Backend"]
        HF[Hugging Face]
        TF[TensorFlow]
        SKL[scikit-learn]
    end

    Browser & Mobile --> NextJS
    NextJS --> API
    API --> Edge
    Edge --> Gemini
    Edge --> LinkedIn
    Actions -->|Trigger| API
    Actions --> Repo
    Repo -->|Deploy| Vercel
    HF --> TF & SKL

    classDef vercel fill:#000,stroke:#fff,color:#fff
    classDef github fill:#24292e,stroke:#fff,color:#fff
    classDef service fill:#4285f4,stroke:#fff,color:#fff
    classDef ml fill:#ff6f00,stroke:#fff,color:#fff

    class NextJS,API,Edge vercel
    class Actions,Repo github
    class Gemini,LinkedIn service
    class HF,TF,SKL ml

flowchart LR
    subgraph Frontend["Frontend"]
        direction TB
        N[Next.js 16] --> R[React 19]
        R --> T[TypeScript 5]
        T --> TW[Tailwind v4]
    end

    subgraph Backend["Backend"]
        direction TB
        PY[Python 3.8+]
        PY --> TF[TensorFlow]
        PY --> SK[scikit-learn]
    end

    subgraph Infra["Infrastructure"]
        direction TB
        V[Vercel Edge]
        GH[GitHub Actions]
        HF[Hugging Face]
    end

    subgraph AI["AI Services"]
        direction TB
        GEM[Google Gemini]
        RAG[RAG Context]
    end

    Frontend <--> Infra
    Backend <--> Infra
    Infra <--> AI

flowchart TB
    subgraph Root["thomas-to-bcheme/"]
        subgraph SRC["src/"]
            APP[app/<br/>Next.js Routes]
            COMP[components/<br/>React UI]
            DATA[data/<br/>RAG Context]
            LIB[lib/<br/>Utilities]
        end

        subgraph BE["backend/"]
            ML[ml_model.py]
            DL[dl_model.py]
            EVAL[evaluation.py]
        end

        subgraph PLUG["plugins/"]
            GIT[git-push<br/>git-commit<br/>git-README]
            AGENT[tto-agent-*<br/>9 specialists]
        end

        subgraph DOCS["system_design_docs/"]
            ARCH[architecture.md]
            APIDOC[api.md]
            DEPLOY[deployment.md]
        end
    end

    APP --> COMP
    COMP --> DATA
    APP --> LIB

sequenceDiagram
    participant C as Client
    participant A as /api/chat
    participant R as RAG Context
    participant G as Google Gemini

    C->>A: POST {messages}
    A->>R: Load AiSystemInformation
    A->>G: Stream request + context
    loop Streaming Response
        G-->>A: Text chunk
        A-->>C: SSE event
    end
    A-->>C: Stream complete

sequenceDiagram
    participant S as GitHub CRON
    participant C as CLI (npm run linkedin)
    participant F as genAI/linkedin-7day/
    participant L as LinkedIn API

    S->>C: post --file {filename}
    C->>F: Read post content
    C->>L: POST ugcPosts
    L-->>C: 201 Created
    C->>F: Update rotation index
    C-->>S: Exit code 0

flowchart LR
    subgraph Git["Git Workflows"]
        GP[git-push]
        GC[git-commit]
        GPA[git-push-agentic]
        GR[git-README]
    end

    subgraph Agents["Specialized Agents"]
        ORCH[orchestrator]
        FE[frontend]
        BE[backend]
        API[api]
        QA[qa]
        ML[ai-ml]
        OPS[ops]
        UX[uiux]
        LI[linkedin]
    end

    subgraph Init["Setup"]
        INIT[tto-init]
        SWE[tto-agent-swe]
    end

    SWE --> Agents
    INIT --> Git

flowchart LR
    subgraph Trigger["Triggers"]
        PUSH[Push to main]
        CRON[CRON Schedule]
        PR[Pull Request]
    end

    subgraph Actions["GitHub Actions"]
        LINT[Lint & Type Check]
        BUILD[Build]
        POST[LinkedIn Post]
    end

    subgraph Deploy["Deployment"]
        PREVIEW[Preview Deploy]
        PROD[Production Deploy]
    end

    PUSH --> LINT --> BUILD --> PROD
    PR --> LINT --> BUILD --> PREVIEW
    CRON --> POST

gantt
    title Implementation Roadmap
    dateFormat YYYY-MM-DD
    axisFormat %b %d

    section Track A: ML Pipeline
    Skill Taxonomy & Salary Parser        :a1, 2026-02-10, 2d
    EDA Notebook (Data Engineering)       :a2, after a1, 3d
    Linear Regression (Ridge)             :a3, after a2, 3d
    Hugging Face Deployment (Gradio)      :a4, after a3, 2d
    Chrome Extension (Manifest V3)        :a5, after a4, 3d

    section Track B: LinkedIn Engine
    Private Repo Setup                    :b1, 2026-02-10, 2d
    Anti-Spam Pipeline (4 Layers)         :b2, after b1, 4d
    Content Migration (20 Posts)          :b3, after b2, 2d
    GitHub Action CRON Workflow           :b4, after b3, 2d
    Claude Code Skills                    :b5, after b4, 3d

    section Track C: README & Docs
    Unified README with Mermaid           :c1, after a3, 2d

flowchart TD
    subgraph Data["Data Foundation"]
        CSV[(jobs_dataset.csv<br/>735 rows)]
        TAX[Skill Taxonomy<br/>DE: 62 keywords<br/>AI: 56 keywords]
        SAL[Salary Parser<br/>Annual/Hourly/Range]
    end

    subgraph EDA["Exploratory Data Analysis"]
        SEG{Skill Classification}
        CSV --> SAL --> SEG
        TAX --> SEG
        SEG -->|No AI skills| PDE["Pure DE<br/>~38 rows"]
        SEG -->|DE + AI skills| HYB["Hybrid DE+AI<br/>~188 rows"]
        SEG -->|No DE skills| PAI["Pure AI<br/>~223 rows"]
    end

    subgraph Models["Dual Ridge Regression"]
        PDE --> MA["Model A<br/>Lower Bound<br/>Ridge Regression"]
        HYB --> MB["Model B<br/>Upper Bound<br/>Ridge Regression"]
        MA --> BAND["Salary Band<br/>$$X - $$Y"]
        MB --> BAND
    end

    subgraph Deploy["Deployment"]
        BAND --> HF["Hugging Face Space<br/>Gradio REST API"]
        HF --> EXT["Chrome Extension<br/>Manifest V3"]
        EXT --> JOB["Job Description<br/>LinkedIn / Indeed"]
    end

    classDef data fill:#1e3a5f,stroke:#4a90d9,color:#fff
    classDef eda fill:#2d5016,stroke:#5cb85c,color:#fff
    classDef model fill:#5c3d1e,stroke:#d9a441,color:#fff
    classDef deploy fill:#3d1e5c,stroke:#9b59b6,color:#fff

    class CSV,TAX,SAL data
    class SEG,PDE,HYB,PAI eda
    class MA,MB,BAND model
    class HF,EXT,JOB deploy

flowchart LR
    subgraph Content["Content Lifecycle"]
        D[drafts/] -->|human review| Q[queue/<br/>not-posted]
        Q -->|pipeline selects| PIPE
        A[archive/<br/>posted] -->|"len >= 20<br/>& 30 days"| Q
    end

    subgraph PIPE["Anti-Spam Pipeline"]
        direction TB
        L4["Layer 4<br/>Frequency Guard<br/>archive >= 20?"]
        L3["Layer 3<br/>Hook Rotation<br/>Hydra Head"]
        L2["Layer 2<br/>Pixel Mutation<br/>Pillow"]
        L1["Layer 1<br/>Zero-Width<br/>Injection"]
        L4 --> L3 --> L2 --> L1
    end

    L1 --> POST[LinkedIn API]
    POST --> A

    classDef queue fill:#1e3a5f,stroke:#4a90d9,color:#fff
    classDef pipe fill:#5c1e1e,stroke:#d94a4a,color:#fff
    classDef post fill:#2d5016,stroke:#5cb85c,color:#fff

    class D,Q,A queue
    class L4,L3,L2,L1 pipe
    class POST post

gitGraph
    commit id: "main"
    branch feature/your-feature
    checkout feature/your-feature
    commit id: "Implement"
    commit id: "Test"
    checkout main
    merge feature/your-feature id: "PR Merged"