The exponential growth of job applications has rendered traditional manual resume screening inefficient, inconsistent, and prone to bias. This research report examines AI-based resume screening and job matching systems that leverage Natural Language Processing, deep learning, and transformer models to automate and optimize talent acquisition. We review state-of-the-art architectures including CNN-Attention for resume topic segmentation, GA-LightGBM and Fuzzy NLP models for human-job matching, and LLM-based systems using GPT-4/GPT-5 embeddings. Empirical results from recent studies show significant performance gains: CNN-Attention achieves 98.42% precision and 99.61% recall in screening, while Fuzzy NLP improves matching accuracy from 25% to 85% and reduces manual review time by 30%. Hybrid NLP + Explainable AI systems demonstrate 90–92% accuracy compared to 70% for manual screening. Key challenges addressed include semantic ambiguity, contextual understanding beyond keywords, bias mitigation, and transformer token limits for long resumes. The report also highlights emerging issues such as LLM self-preferencing and the need for transparency in AI-driven hiring decisions. Findings indicate that AI-based systems not only accelerate shortlisting by over 50% but also improve interview rates, with tailored applications securing interviews at more than double the rate of generic submissions. Future research directions point toward graph neural networks, Model Context Protocol integration, and ontology-based skill matching for fairer, more interpretable, and scalable recruitment platforms.

Keywords: Resume Screening, Job Matching, Natural Language Processing, Transformer Models, Fuzzy Logic, Bias Mitigation, Explainable AI, Talent Acquisition