README.md

metadata

title: Assistant-Tiya
emoji: 🤖
colorFrom: blue
colorTo: purple
sdk: streamlit
sdk_version: 1.28.1
app_file: app.py
pinned: false

AI Resume Assistant

A Streamlit-based LLM assistant that helps you tailor your resume and cover letter based on job descriptions.

Features

1. Resume Finder - Automatically selects the most relevant portfolio (AI or Data Science) based on job description
2. Cover Letter Generator - Creates customized cover letters using your selected resume with AI-powered web search for company research
3. General Query - Answers questions about your application with relevant context

Setup

1. Install Dependencies

pip install -r requirements.txt

2. Configure API Keys

Create a .env file in the project directory:

OPENROUTER_API_KEY=your_openrouter_api_key_here
OPENAI_API_KEY=your_openai_api_key_here

Or provide API keys through the Streamlit UI sidebar.

3. Ensure Portfolio Files Exist

Make sure you have:

• AI_portfolio.md
• DS_portfolio.md

in the same directory as app.py

4. Create README.md File (For HuggingFace Spaces)

Create a README.md file in your project root with: ```yaml

title: Assistant-Tiya emoji: 🤖 colorFrom: blue colorTo: purple sdk: streamlit sdk_version: "1.28.1" app_file: app.py pinned: false

This configuration file is **required** for HuggingFace Spaces deployment. It tells HuggingFace:
- The space name and appearance
- That it's a Streamlit app
- Which file to run (`app.py`)

### 5. Run the Application
```bash
streamlit run app.py

How It Works

Input Section

• Job Description (Required) - Paste the job posting
• Resume Finder (Optional) - Check to find the best matching resume
• Cover Letter (Optional) - Check to generate a cover letter with web search
• Length of Resume (Optional) - Choose response length (short/medium/long)
• Select Resume (Optional) - Choose AI_portfolio or DS_portfolio
• Entry Query (Optional) - Ask specific questions (max 5000 characters)

Processing Logic

The system categorizes inputs based on your selections:

Resume Finder	Cover Letter	Select Resume	Action
✓	✗	No Select	Resume Finder - Picks best portfolio
✗	✓	Resume_P/Dss	Cover Letter - Generates with web search context
✗	✗	Resume_P/Dss	General Query - Answers your question
Any other combination			Retry - Validation error

Output

Each category produces:

1. Resume Finder Output

   • Selected Resume: Resume_P or Resume_Dss
   • Reasoning: 25-30 words explaining why

2. Cover Letter Output

   • Professional cover letter (<250 words)
   • Starts with "Dear Hiring Manager,"
   • Ends with "Best,\nDhanvanth Voona" (signature on separate line)
   • No em dashes or corporate jargon
   • Personalized with company-specific research via web search

3. General Query Output

   • Natural human-like response
   • Tailored to specified word count
   • Ready to use in applications

Technologies Used

• Frontend: Streamlit
• LLM Models:
  • Resume Finder: openai/gpt-oss-safeguard-20b (via OpenRouter)
  • Cover Letter & General Query: gpt-5-mini-2025-08-07 (OpenAI)
  • Cover Letter Context (Web Search): perplexity/sonar (via OpenRouter)
• APIs: OpenAI, OpenRouter (with Perplexity web search capability)
• PDF Generation: ReportLab
• Timezone Support: Pytz (EST timestamps)

Key Features

✨ Auto-Generated Company Research - Uses Perplexity/Sonar to search the web for company information
✨ Two-Stage Cover Letter Generation - First generates company-specific context, then writes personalized letter
✨ PDF Download - Export cover letters as PDFs with EST timestamps
✨ Edit & Save - Modify responses and save changes before downloading
✨ Multi-Portfolio Support - Switch between AI and Data Science portfolios
✨ No Generic Templates - All outputs are authentic and personalized

Notes

• All LLM outputs are without preamble for direct usage
• Cover letters have no em dashes as per specifications
• General queries are kept under word limits specified
• Responses are formatted to be immediately usable in applications
• Web search for cover letter context requires OpenRouter API key (Perplexity/Sonar model)

---

Created by: Dhanvanth Voona