tariwecode/Automating-CV-extraction-and-analysis-using-NLP-and-document-AI

A lightweight, AI-powered recruitment screening tool built for small and medium-sized enterprises (SMEs) in Zimbabwe. EasyHire parses CVs in multiple formats, extracts and matches skills to job descriptions, ranks candidates transparently, and performs basic certificate authenticity checks — all from a single Streamlit dashboard. # EasyHire — CV Extraction & Analysis System > Automating CV Extraction and Analysis Using NLP and Document AI A lightweight, AI-powered recruitment screening tool built for small and medium-sized enterprises (SMEs) in Zimbabwe. EasyHire parses CVs in multiple formats, extracts and matches skills to job descriptions, ranks candidates transparently, and performs basic certificate authenticity checks — all from a single Streamlit dashboard. --- ## Problem SMEs in Zimbabwe typically lack dedicated HR departments. Hiring is handled by owners or administrators who receive CVs via WhatsApp, email, and paper in inconsistent formats (scanned images, PDFs, Word documents). Manual screening is slow, inconsistent, and prone to bias. Existing Applicant Tracking Systems are expensive, bandwidth-heavy, and impractical for lean teams. ## Solution EasyHire automates the first-pass shortlisting stage of recruitment by combining document parsing, semantic search, and generative AI into a single affordable workflow that runs on ordinary hardware. --- ## Features - **Multi-format CV parsing** — Extracts text from PDF, DOCX, and plain-text files using PyMuPDF and python-docx. - **Semantic embeddings & vector search** — Converts CV content into vector embeddings (BGE-Small) and indexes them with FAISS for fast similarity retrieval. - **RAG-powered analysis** — Retrieves the most relevant CV chunks and feeds them to Google Gemini for context-aware candidate evaluation. - **Skill extraction & matching** — Identifies required skills from job descriptions, matches them against CV content, and highlights gaps. - **Match percentage scoring** — Produces a numerical suitability score combining keyword matches, semantic relevance, and LLM scoring. - **ATS-style candidate reports** — Generates structured feedback: profile summary, strengths, weaknesses, and missing skills. - **Candidate ranking dashboard** — Displays ranked shortlists with filters for experience, match percentage, and de …