Skip to content

About

7-agent LLM orchestration system — parallel execution, scam detection, opportunity scoring & auto-reply drafting via AXIS orchestrator, FastAPI, and React dashboard.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Latest commit

 

History

4 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 

Repository files navigation

⚡ Job Intel AI — The Ultimate Career Command Center

Job Intel AI is a cutting-edge, real-time automation system designed for elite Students. It transforms the chaotic noise of job alerts and freelance inquiries into a prioritized, high-signal stream of actionable opportunities, using a Multi-Agent LLM Architecture — a coordinated fleet of 7 specialized AI agents — to detect scams, score opportunities, and draft ready-to-send professional responses.


🏗️ Multi-Agent System Architecture

The system is composed of 7 specialized AI agents, each owning a distinct responsibility. AXIS, the Orchestrator Agent, coordinates all agents, manages task delegation, and assembles the final intelligence report.

graph TD
    subgraph "Detection Layer (Browser)"
        A[Distill.io / Browser Extension] -->|Monitors| B(11+ High-Value Platforms)
        A -->|Webhook POST| C[FastAPI Ingestion Endpoint]
    end

    subgraph "Ingestion Layer (Backend)"
        D[Gmail / alert@distill.io] -->|IMAP Polling| E[Email Worker Service]
        E -->|Internal API Call| C
        C -->|Saves to| F[(SQLite Database)]
    end

    subgraph "🤖 Multi-Agent Intelligence Layer"
        F -->|Trigger| AXIS

        AXIS[🧠 AXIS — Orchestrator Agent]

        AXIS -->|Fetch context| ECHO[💾 ECHO — Memory Agent]
        AXIS -->|Parallel dispatch| SHIELD[🛡️ SHIELD — Scam Detection Agent]
        AXIS -->|Parallel dispatch| LENS[📊 LENS — Scoring Agent]

        ECHO -->|User prefs + history| AXIS
        SHIELD -->|Risk report + red flags| AXIS
        LENS -->|Score 1-10 + reasoning| AXIS

        AXIS -->|If score > 4| INK[✉️ INK — Reply Drafting Agent]
        AXIS -->|If score > 7| FORGE[🚀 FORGE — Readiness Agent]

        INK -->|Draft reply| AXIS
        FORGE -->|Profile checklist| AXIS

        AXIS -->|Store results| ECHO
        AXIS -->|Structured Report| OUT{📋 Final Intelligence Report}
    end

    subgraph "Action Layer (UI)"
        OUT --> M[React Dashboard]
        M -->|User Action| N[Copy Reply / Mark Done / Export]
    end
Loading

🤖 Agent Roster

Agent Codename Role Responsibility
🧠 Orchestrator Agent AXIS Master Controller Routes tasks, manages agent lifecycle, assembles the final report
📡 Ingestion Agent PULSE Data Collector Normalizes raw webhook/email data into a structured schema
🛡️ Scam Detection Agent SHIELD Risk Auditor Scans for red flags, suspicious patterns, and blacklisted behaviors
📊 Scoring Agent LENS Opportunity Analyst Scores leads 1–10 based on profile alignment, value & strategy
✉️ Reply Drafting Agent INK Communications Expert Generates personalized, platform-specific professional responses
🚀 Readiness Agent FORGE Career Coach Audits LinkedIn/GitHub, suggests improvements for high-value leads
💾 Memory Agent ECHO Context Keeper Persists user preferences, learns patterns across sessions

Agent Tier Strategy: AXIS handles complex orchestration reasoning at the highest capability tier. SHIELD, LENS, and INK handle nuanced analysis and writing at the mid tier. PULSE, FORGE, and ECHO handle fast, lightweight utility tasks at the speed tier — keeping the system efficient and cost-effective.


🧪 Multi-Agent Intelligence Flow

sequenceDiagram
    participant W as Distill.io Webhook
    participant API as FastAPI Backend
    participant AXIS as 🧠 AXIS (Orchestrator)
    participant ECHO as 💾 ECHO (Memory)
    participant SHIELD as 🛡️ SHIELD (Scam)
    participant LENS as 📊 LENS (Scoring)
    participant INK as ✉️ INK (Reply)
    participant FORGE as 🚀 FORGE (Readiness)
    participant DB as SQLite

    W->>API: Raw Change Notification
    API->>DB: Store Raw Lead (Pending)
    API->>AXIS: Dispatch lead for analysis

    AXIS->>ECHO: Fetch user context & preferences
    ECHO-->>AXIS: Historical patterns + user profile

    par Parallel Execution
        AXIS->>SHIELD: Analyze for scams & red flags
        AXIS->>LENS: Score opportunity 1-10
    end

    SHIELD-->>AXIS: Risk level + red flag list
    LENS-->>AXIS: Score + detailed reasoning

    alt Score > 4 (Viable Lead)
        AXIS->>INK: Draft professional reply
        INK-->>AXIS: Platform-specific response draft
    end

    alt Score > 7 (High-Value Lead)
        AXIS->>FORGE: Run profile readiness check
        FORGE-->>AXIS: Profile gaps + improvement checklist
    end

    AXIS->>ECHO: Store results & update patterns
    AXIS->>API: Return Final Intelligence Report
    API->>DB: Update lead with full analysis
    API-->>User: Real-time Dashboard Update
Loading

🧠 Core Features

This is not just another job tracker. It's a fleet of intelligent agents working 24/7 in parallel.

1. 🛡️ SHIELD — Scam Detection Agent

Every opportunity is automatically audited for potential "Red Flags."

  • Detection criteria: Unusual hourly rates, requests for free work, generic templates, high-pressure tactics, or known scam patterns.
  • Output: A clear, bulleted list of warnings with a Low / Medium / High risk rating.

2. 📊 LENS — Scoring Agent

Not all leads are created equal. Scoring logic is based on:

  • Alignment: How well the role fits an AI Engineer profile.
  • Value: Estimated contract value or strategic importance.
  • Reasoning: A detailed explanation of the priority score (1–10).

3. ✉️ INK — Reply Drafting Agent

Stop staring at a blank screen. INK generates professional, contextual replies for every lead.

  • Customized: References specific project details from the alert.
  • Platform-aware: Adapts tone and format for LinkedIn, Upwork, Email, etc.
  • Copy-Paste Ready: One-click copying directly from the dashboard.

4. 🚀 FORGE — Readiness Agent

The system proactively ensures you are ready for the opportunity:

  • Checklist: "Is your LinkedIn/GitHub/Landing Page optimized for this specific client?"
  • Guidance: For high-value leads (score > 7), FORGE suggests targeted improvements before you hit send.

5. 💾 ECHO — Memory Agent

The system learns and improves over time:

  • Preference Learning: Remembers your preferences (e.g., preferred rate range, avoided client types).
  • Pattern Recognition: Identifies which opportunity types convert best for your profile.
  • Cross-Session Context: Every new lead is analyzed with the full weight of your history.

🔄 Agent Communication Protocol

Agents communicate via a structured internal message schema:

{
  "task_id": "uuid-v4",
  "source_agent": "AXIS",
  "target_agent": "SHIELD",
  "payload": {
    "lead_content": "Raw opportunity text...",
    "platform": "upwork",
    "timestamp": "2025-07-10T14:30:00Z"
  },
  "context": {
    "user_profile": "AI Engineer, 3 years exp...",
    "historical_patterns": ["prefers async roles", "avoids < $50/hr"]
  }
}

AXIS collects all agent responses and assembles the Final Intelligence Report:

{
  "lead_id": "uuid-v4",
  "platform": "upwork",
  "analysis": {
    "agent_shield": {
      "risk_level": "Low",
      "red_flags": []
    },
    "agent_lens": {
      "opportunity_score": 8.4,
      "score_reasoning": "Strong alignment with AI engineering profile..."
    },
    "agent_ink": {
      "suggested_reply": "Hi [Name], I'd love to discuss your ML pipeline project..."
    },
    "agent_forge": {
      "readiness_checklist": [
        "✅ LinkedIn profile is up to date",
        "⚠️  Add recent RAG project to GitHub",
        "✅ Portfolio landing page is live"
      ]
    }
  }
}

🏗️ Technical Implementation

Backend: The Engine

  • FastAPI: Asynchronous, high-performance API routing.
  • IMAP Worker: Dedicated email_worker.py running every 20 seconds for zero-latency detection.
  • SQLAlchemy: Robust ORM for persistent data management.
  • LLM SDK: Powers the full agent fleet — AXIS, SHIELD, LENS, INK, FORGE, PULSE, and ECHO.
  • Async Agent Dispatcher: Parallel agent execution via asyncio.gather() for maximum speed.

Frontend: The Command Center

  • React.js (Vite): Lightning-fast development and optimized build.
  • Mermaid.js Integration: For architectural transparency.
  • Custom CSS: A high-end, professional "Command Center" aesthetic with dark-espresso accents and parchment-cream backgrounds.

Agent Execution Model

AXIS (Orchestrator)
├── [Sequential]  ECHO    → Fetch context
├── [Parallel]    SHIELD  → Risk analysis
├── [Parallel]    LENS    → Opportunity score
├── [Conditional] INK     → Draft reply   (if score > 4)
├── [Conditional] FORGE   → Profile check (if score > 7)
└── [Sequential]  ECHO    → Store results

📡 Supported Platforms (The Big 11)

  • Networking: LinkedIn, Bluesky, DEV Community
  • Freelance: Upwork, Fiverr, Freelancer
  • Job Boards: ZipRecruiter, SimplyHired, Glassdoor
  • Specialized: Kaggle (Competitions), Google Scholar (Research alerts)

🚀 One-Command Launch

Launch the entire ecosystem (Backend, Frontend, and Worker) in seconds:

bash start.sh

🗺️ Multi-Agent Upgrade Roadmap

Phase Agent Status
✅ Phase 1 Single-agent analysis (legacy) Complete
🔄 Phase 2 AXIS + SHIELD + LENS Complete
🔜 Phase 3 INK + FORGE In Progress
🔜 Phase 4 ECHO — cross-session learning Planned
🔜 Phase 5 DEAL — Negotiation Agent (counter-offer advisor) Planned

About

7-agent LLM orchestration system — parallel execution, scam detection, opportunity scoring & auto-reply drafting via AXIS orchestrator, FastAPI, and React dashboard.

Resources

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages