ResumeIQ: Professional Multi-Agent Resume Optimization

ResumeIQ is an advanced AI-driven suite designed to bridge the gap between talented professionals and complex Applicant Tracking Systems (ATS). By orchestrating a sequential workforce of four specialized agents, ResumeIQ transforms raw, unoptimized resumes into high-impact professional documents.

🌉 The Multi-Agent Pipeline

The core of ResumeIQ is a Stateful Language Graph that maintains context as it passes your information through four distinct stages of intelligence:

graph LR
    User([PDF Upload]) --> Parser{{1. Parser Agent}}
    Parser --> ATS{{2. ATS Evaluator}}
    ATS --> Coach{{3. Improvement Coach}}
    Coach --> Rewrite{{4. Rewrite Agent}}
    Rewrite --> Result([Final Optimized Draft])
    
    style User fill:#f9f,stroke:#333,stroke-width:2px
    style Result fill:#00ff00,stroke:#333,stroke-width:2px

Loading

🧠 Agent Responsibilities

1. Parser Agent: Strategically deconstructs the PDF into structured data (Skills, Education, Experience) ensuring no vital info is lost in transit.
2. ATS Evaluator (Mathematical): Applies a cold, objective mathematical rubric (starting from base 40) to calculate your resume's "vulnerability score" based on keywords and metrics.
3. Improvement Coach: Identifies weak bullet points and coaches you on quantifying impact using specific industry action verbs.
4. Rewrite Agent: Synthesizes all criticisms and keywords into a final, polished draft that is ready for submission.

🔑 Bring Your Own Key (BYOK)

To ensure maximum security and privacy, ResumeIQ does not store API keys.

• When running locally or on the deployed version, you must provide your own Groq API Key (or compatible vendor key).
• All processing happens in-memory and is purged immediately after your session ends.

---

💻 Local Setup & Development

Follow these steps to get ResumeIQ running on your local machine:

1. Environment Preparation

Create and activate a virtual environment to keep your dependencies isolated:

Mac/Linux:

python3 -m venv .venv
source .venv/bin/activate

Windows:

python -m venv .venv
.venv\Scripts\activate

2. Install Dependencies

pip install -r requirements.txt

3. Run the Application

Start the Streamlit web server:

streamlit run multi_agent_system.py

🛑 How to stop & Clean up

• Stop the App: Press Ctrl + C in your terminal to shut down the Streamlit server.
• Deactivate Environment: Once you are done, simply run:

  deactivate

---

🎓 Key Learnings

Building ResumeIQ provided valuable insights into the intersection of Generative AI and production-level web engineering:

1. Multi-Agent Orchestration: I learned how to use LangGraph to manage complex state transitions between specialized AI agents, ensuring that analytical findings from one agent are reliably passed to the creative rewrite phase.
2. Streaming & User Experience: Implementing .stream() instead of .invoke() taught me how to bridge the gap between long-running AI tasks and responsive UIs, providing users with immediate visual feedback during complex background calculations.
3. Prompt Engineering & Rubrics: I discovered that LLM "politeness" can be a blocker for analytical tasks. Moving from generic scoring to a mathematical point-based rubric was essential for generating authentic, variarable ATS scores.
4. UI/UX Stability in Streamlit: Navigating Streamlit's "unmounting" behavior was a key learning. By using session_state to strategically hide and show widgets, I solved persistent UI glitches like duplicate icons and doubling labels.
5. Secure AI Architectures (BYOK): Implementing a Bring Your Own Key model highlighted the importance of privacy-first design in AI tools, ensuring that sensitive credentials never leave the user's browser session.