How aashish2502/hire-helper works
Project type and purpose
• This is a full-stack AI-powered screening assistant built with Streamlit (frontend) and Python (backend utilities).
• Purpose: allow HR users to upload resumes, generate embeddings, store/retrieve them, and drive a conversational AI assistant via external LLM APIs (Groq API) with monitoring (LangSmith).
File structure
• Top-level:
• app/app.py → Streamlit-driven UI entry point
• app/utils/embeddings.py → embedding generation logic (Hugging Face + FAISS integration)
• app/utils/history.py → session or conversation history management
• requirements.txt: third-party libraries (Streamlit, Transformers, FAISS, Groq, LangSmith)
README for deeper insights
• Frontend: Streamlit
• Conversational AI: Groq API for LLM calls; LangSmith for tracing/monitoring
• Embeddings & Retrieval: Hugging Face Transformers + FAISS vector store
• Backend pipeline: preprocess → embed → store → retrieve → augment LLM context → respond
• Monitoring: LangSmith logs queries, latencies, reliabilities