CareerLens AI: An Intelligent Multi-Tier Resume Analysis and Job Matching Platform Using Semantic Embeddings and Generative AI

This paper presented CareerLens AI, a full-stack AI-powered career intelligence platform that addresses the systematic disadvantage faced by job seekers in ATS-dominated recruitment environments. The system integrates NLP-based resume parsing (spaCy), multi-dimensional ATS scoring, semantic job matching via sentence-transformer embeddings, and Google Gemini 1.5 Flash generative feedback into a unified, production-deployable three-tier microservice architecture. Evaluation demonstrated that the semantic embedding-based matching approach (F1 = 0.76) significantly outperforms TF-IDF (F1 = 0.59) and keyword-only (F1 = 0.58) baselines. A user study of 15 participants showed a 79% relative improvement in ATS screening pass rates after applying platform recommendations. Gemini-generated recruiter feedback received a mean helpfulness rating of 4.1/5.0 from 20 evaluators. The platform's microservice architecture ensures independent scalability of each tier, and its dual-storage design provides operational resilience across deployment environments. CareerLens AI represents a meaningful step toward equalising access to career intelligence tools that have traditionally been available only through expensive professional services. Source code and evaluation data will be made available upon publication.

[1]   Resume Go, "The Truth About The Applicant Tracking System," Resume Go Research, 2018. [Online]. Available: https://www.resumego.net/research/ats-resume-review/ [2] LinkedIn, "Future of Recruiting 2022," LinkedIn Talent Solutions, 2022. [Online]Available: https://business.linkedin.com/talent-solutions/future-of-recruiting [3] Jobscan, "Job Seeker Nation Report," Jobscan Research, 2020. [Online]. Available: https://www.jobscan.co/blog/job-seeker-nation/ [4]  D. Celik and A. Elci, "An Ontology-Based Candidate Recommendation System Using Semantic Resume Parsing," in Proc. Int. Conf. on Semantic Technologies, 2013, pp. 1–10. [5]   Y. Lin, P. Xu, and Z. Zhang, "BERT-Based Resume Information Extraction and Job Matching," in Proc. 2020 IEEE Int. Conf. on Big Data, IEEE, 2020, pp. 2397–2404. [6]  M. Honnibal, I. Montani, S. Van Landeghem, and A. Boyd, "spaCy: Industrial-strength Natural Language Processing in Python," Zenodo, 2020. https://doi.org/10.5281/zenodo.1212303 [7] G. Salton and C. Buckley, "Term-weighting approaches in automatic text retrieval," Information Processing & Management, vol. 24, no. 5, pp. 513–523, 1988. [8]  N. Reimers and I. Gurevych, "Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks," in Proc. EMNLP, 2019, pp. 3982–3992. [9]   C. Qin et al., "Enhancing Person-Job Fit for Talent Recruitment: An Ability-aware Neural Network Approach," in Proc. 41st Int. ACM SIGIR Conf., 2018, pp. 25–34. [10]   OpenAI, "GPT-4 Technical Report," arXiv preprint arXiv:2303.08774, 2023.