RecruitAI: An Intelligent Resume Screening and Analysis Platform Based on BERT and FAISS

RecruitAI is an open-source AI-driven resume screening and analysis platform that integrates NLP technologies including BERT semantic similarity, TF-IDF text analysis, and FAISS vector search. It addresses pain points like low efficiency and subjective bias in traditional recruitment screening, enabling functions such as automatic candidate ranking, skill gap analysis, and semantic search. The platform supports multi-format resume input, can be deployed locally or in the cloud, and data privacy is controlled by users, providing enterprises with production-level intelligent analysis capabilities.

Traditional manual resume screening is inefficient: large enterprises receive an average of 250 resumes per position, with initial screening time accounting for over 40% of the recruitment cycle, and are prone to missing excellent candidates due to subjective bias. With the development of NLP and deep learning technologies, AI-driven screening systems have become a solution. As an open-source project, RecruitAI integrates cutting-edge technologies into a Streamlit application to improve screening efficiency and accuracy.

Hybrid Scoring Model: Combines BERT semantic similarity (50% weight, understands contextual relevance), TF-IDF similarity (20%, supplements domain term recognition), skill matching degree (20%, extracts keywords to calculate matching rate), and experience years (10%, parses work experience); FAISS Semantic Search: Converts resumes into vectors for storage, supports natural language search, and returns results in milliseconds; NLP Skill Extraction: Uses spaCy to extract skills and generate skill gap reports.

• Multi-format Upload: Supports PDF single files, ZIP batch compression packages, and CSV structured data;
• Resume Evaluation: Provides resume strength score (quality integrity) and recruitment readiness classification (Ready/Potential/Needs Work);
• Visual Dashboard: Skill distribution radar chart, candidate scatter plot, and intelligent skill analysis;
• Result Export: Supports CSV export and can be integrated into ATS systems.

• Technical Position Recruitment: Efficiently processes large numbers of technical resumes and identifies tech stack matching degree;
• Campus Recruitment: Batch processes resumes and builds talent reserves via strength scores;
• Headhunting Services: Uses semantic search to quickly match job requirements and skill gap analysis to assist recommendations;
• Process Optimization: Accumulates historical data to optimize job descriptions and recruitment processes.

• Local Deployment: Clone the repository → Install dependencies → Download spaCy models → Launch Streamlit application;
• Docker Deployment: Build image → Run container to ensure environment consistency;
• Streamlit Cloud: Push code to GitHub, deploy to the cloud after configuration, and make it accessible via public network.

Current Limitations: Mainly supports English resumes, with insufficient quantification of soft skills and cultural fit; Future Plans: Multi-language support (XLM-RoBERTa), LLM-generated interview questions, automatic JD skill extraction, candidate pipeline tracking, and email integration.

RecruitAI lowers the threshold for using intelligent recruitment systems and helps enterprises improve screening efficiency, but it should be used as an auxiliary tool rather than replacing human judgment. It is recommended that enterprises start with auxiliary screening to build trust in AI decisions, be alert to algorithmic bias, regularly review recommendation quality, and ensure that technology serves the goal of talent discovery.