⚡ Shark AI Dashboard

A Black-Gold Powered AI System for Real-Time Monitoring, Dataset Generation & LLM Workflows

---

✨ Overview

Shark AI Dashboard is a full-stack AI control panel designed for high-performance workflows:

• ⚙️ Real-time system monitoring (CPU, Memory, Charts, GPU-ready)
• 📚 Intelligent text chunking & QA-pair generation
• 🔥 Streaming algorithm execution (SSE)
• 🧠 Customizable LLM pipeline with parameters & file upload
• 💎 Black-Gold premium UI with motion effects & data visualizations
• 📡 Key usage burn-down tracker (ChatAnywhere API cost tracking)
• 🕸 Interactive Model Graph & Heatmaps

This dashboard is built not just as a tool —
but as a control center for AI developers and researchers.

---

🚀 Core Features

🧠 1. QA Pair Generation Pipeline

• Upload .txt files
• Dynamic chunk splitting
• LLM inference (GPT-5 series / custom API models)
• Realtime progress (%), per-chunk timing
• Auto-formatted JSONL export (Alpaca style)

📊 2. Real-Time Monitoring

• Dynamic CPU usage gauge
• Memory streaming line chart (rolling window)
• Task statistics bar chart
• Heatmap / spectrum data visualization
• Configurable refresh interval

💡 3. System Visualization

• Network graph of model workflow
• Black-gold lighting, shadow, and motion UI
• Adaptive, responsive layout

⚡ 4. Streaming Algorithm Execution (SSE)

• POST to queue workload
• GET SSE for live logs + tokens
• Frontend burn-down animation & status console

🔐 5. API Key Usage Dashboard

• Cost aggregation from ChatAnywhere (/v1/query/usage_details)
• Remaining balance percentage
• Burn-down bar
• History breakdown

---

🛠 Tech Stack

Frontend

• React (Vite)
• TailwindCSS + Custom Black-Gold Theme
• ECharts (line/gauge/bar/heatmap/network)
• Lucide Icons
• SSE streaming support

Backend

• FastAPI
• psutil (system stats)
• OpenAI-compatible API (ChatAnywhere)
• Chunking engine for document processing
• SSE log streaming endpoint

---

📦 Project Structure

frontend/ ├─ src/ │ ├─ components/ │ ├─ pages/ │ ├─ charts/ │ └─ router/ backend/ ├─ api/ ├─ monitor/ └─ qa_generation/

---

🎨 Preview

The interface features high-contrast neon gold accents, animated gradients, glowing charts, and a clean industrial layout.

(You can add screenshots here)

---

🧩 Why This Exists

Building datasets, monitoring LLM pipelines, and debugging AI workloads is messy.
This dashboard brings all essential tools into one coherent, powerful, elegant UI.

If you're training models, generating QA datasets, or running local AI systems —
this becomes your cockpit.

---

⭐ Future Enhancements

• Multi-key usage dashboard
• GPU temperature + VRAM monitoring
• Multi-agent workflow designer
• Real-time token usage meter
• Plugin system for custom algorithms

---

🤝 Contributions

PRs welcome.
If you build a plugin / new chart / new algorithm panel — send it in!

---

📜 License

MIT