How ri-100/resumexpert works
Main Components
a) User Interface Layer:
• Streamlit Web Application (represented by app/main.py)
• Input Components: Job Description Upload
• Output Components: Resume Match Results, ATS Score, Cold Email
b) Application Layer:
• Core Application Logic (app/app.py)
• Portfolio Management (app/portfolio.py)
• LangChain Integration (app/chains.py)
• Utility Functions (app/utils.py)
c) Data Storage Layer:
• ChromaDB Vector Store (vectorstore directory)
• Resume Portfolio Storage (app/resource/my_portfolio.csv)
d) External Services:
• LangChain Services
• Language Models (likely OpenAI, based on the tech stack)
Application Logic → Cold Email Generation → User Interface
[User] → [Streamlit Interface]
• ↓
[Application Core]
• ↙ ↓ ↘
[Portfolio] [Chains] [Utils]
• ↓ ↓
[ChromaDB] [Language Model]
Key Architectural Patterns to Highlight
• Vector-based Document Matching Architecture
• Service-Oriented Integration with LangChain
• MVC-like pattern with Streamlit as View, app.py as Controller, and portfolio.py as Model
Technical Details
• ChromaDB vector store configuration
• LangChain integration points
• Resume processing pipeline
• ATS scoring mechanism
• Email generation workflow
Implementation Notes
• The separation between the web interface and core logic
• The vector store's role in resume matching
• The integration with LangChain for NLP tasks
• The data flow for ATS score calculation
• The cold email generation pipeline