SmartATS: A Lightweight, Fully Offline Rule-Based Framework for Resume Analysis and ATS Compatibility Scoring

SmartATS is a lightweight, fully offline, rule-based framework for resume analysis and Applicant Tracking System (ATS) compatibility scoring. The framework addresses the limitations of cloud-based and Large Language Model (LLM)-powered resume analyzers, including dependency on internet connectivity, external API keys, recurring API costs, non-deterministic outputs, and privacy concerns associated with transmitting personal resume data to third-party services. SmartATS performs resume analysis entirely on the user's local machine using deterministic rule-based techniques. The framework provides ATS compatibility scoring, resume section completeness detection, contact-information validation, role-specific keyword and skills-gap analysis, action-verb detection, quantified-achievement analysis, resume-length evaluation, and actionable improvement suggestions. The system uses a transparent six-factor weighted scoring model consisting of section completeness, contact-information completeness, role-specific keyword matching, impact and action verbs, quantified achievements, and resume length. The current implementation supports seven career tracks: Software Engineering, Data Analysis and Data Science, SOC Analysis and Cybersecurity, Digital Marketing, Sales and Business Development, Automotive Sales and Marketing, and Content Creation and Social Media. SmartATS is implemented using Python and Streamlit, with pdfplumber and python-docx for resume parsing and Plotly for visualization. The core analysis engine requires no machine-learning model, external API, or network connection. The paper presents the system architecture, scoring methodology, implementation details, computational complexity, role-specific keyword taxonomy, comparative analysis against a representative Gemini-based resume analyzer, reproducibility evaluation, before-and-after case studies, component sensitivity analysis, limitations, privacy considerations, and future research directions. The results demonstrate that a deterministic offline resume analyzer can provide fast, transparent, reproducible, privacy-preserving, and zero-marginal-cost ATS-oriented feedback. The study also identifies the limitations of rule-based keyword matching and proposes a hybrid architecture combining fast local analysis with an optional generative AI-based review for future development.