Interview IQ³: An AI-Powered Intelligent Hiring Platform for Resume Analysis, Adaptive Interview Practice and Recruiter Workflow Management

Abstract

Recruitment is still slow and uneven: recruiters read large numbers of resumes, applicant tracking systems filter by keywords, and candidates receive little feedback after an interview. Existing tools cover one stage of hiring each, so recruiters and candidates must use separate platforms. This paper presents Interview IQ³, an artificial intelligence (AI) based hiring platform that joins resume analysis, adaptive interview practice and recruiter workflow management in one web application. A Llama 3 language model accessed through an API reads the text of an uploaded resume and returns an applicant-tracking-system (ATS) score, missing keywords, strengths, weaknesses and suggested roles. An adaptive interview generates questions from the resume, the target role, the experience level and a company-specific mode, accepts answers by text or voice, and evaluates them for technical accuracy, confidence, fluency, grammar and relevance with BERT and Sentence Transformer models. Companies create jobs and assessment rounds, and recruiters compare AI-generated reports and shortlist candidates. The system is built with React, Node.js, Express, MongoDB and Firebase. The architecture, scoring rule and prototype are described, and a reproducible evaluation plan is given.

Introduction

The text presents Interview IQ³, an integrated AI-based recruitment and interview preparation platform designed to improve the hiring process for both recruiters and job candidates. Traditional recruitment involves manually screening large numbers of resumes, scheduling interviews, and comparing candidates, while job seekers often prepare using separate resources and receive limited feedback after interviews. These disconnected processes make recruitment time-consuming and provide little transparency regarding selection decisions.

To overcome these challenges, Interview IQ³ combines resume analysis, intelligent job matching, adaptive interview preparation, answer evaluation, and recruiter management within a single web-based platform. It uses Artificial Intelligence and Natural Language Processing (NLP) to understand candidate profiles, generate relevant interview questions, evaluate responses, and produce detailed performance reports.

The literature survey reviews 12 research studies covering AI-based recruitment, candidate-job matching, algorithmic hiring bias, language understanding, sentence embeddings, question generation, and foundation language models. Previous studies demonstrate the potential of technologies such as BERT, Sentence-BERT, LLaMA, and large language models for semantic text analysis and automated question generation. However, they generally address individual recruitment activities rather than providing a complete platform that connects candidates and recruiters.

The research identifies four major gaps:

• Separate recruitment tools: Resume screening, interview preparation, and recruiter management operate independently without sharing candidate information.

• Keyword-based screening: Traditional Applicant Tracking Systems (ATS) may overlook qualified candidates whose skills are expressed using different terminology.

• Lack of transparent feedback: Candidates and recruiters often receive scores without sufficient explanations or improvement recommendations.

• Disconnected preparation and hiring: Generic interview preparation tools are not aligned with the actual job requirements and company-specific recruitment processes.

The study establishes six main objectives:

1. Develop secure registration and login with role-based access for administrators, companies, recruiters, and candidates.

2. Analyse uploaded resumes and generate ATS scores, missing keywords, strengths, weaknesses, and suitable job-role suggestions.

3. Generate personalized interview questions based on the candidate's resume, job role, experience level, and company requirements.

4. Evaluate interview responses submitted through text or voice and provide scores, feedback, and improvement suggestions.

5. Enable companies to create job openings and assessment rounds while allowing recruiters to compare, shortlist, and evaluate candidates.

6. Maintain historical records of resumes, interviews, and evaluation reports for future reference.

The proposed methodology follows a three-tier architecture consisting of a React.js frontend, a Node.js and Express.js backend, and a MongoDB Atlas database. Tailwind CSS is used for interface design, while Firebase Authentication supports user authentication. AI models are integrated into the backend to perform resume analysis, interview question generation, and response evaluation.

The platform supports four types of users:

• Administrator: Verifies company accounts, manages users and recruiters, and monitors platform activities.

• Company: Creates company profiles, publishes job openings, defines eligibility criteria, and configures interview rounds.

• Recruiter: Schedules interviews, reviews AI-generated reports, compares candidates, shortlists applicants, and updates recruitment outcomes.

• Candidate: Uploads resumes, participates in practice and official interviews, and accesses performance feedback and historical reports.

Security is implemented through role-based access control, bcrypt password hashing, JSON Web Tokens (JWT), and Firebase-based Google sign-in. The backend provides REST APIs for managing users, companies, recruiters, candidates, resumes, and interviews. The frontend is deployed using Netlify, while the backend is hosted on Render.

Conclusion

Interview IQ³ combines AI resume analysis, adaptive interview practice and recruiter workflow management in one web platform, functions that existing tools provide separately. Role-based access, a stored history and a simple, stated rule for the overall score keep the platform understandable to both candidates and recruiters. Future work includes the quantitative evaluation described above, grounding questions and feedback in a curated question bank by retrieval-augmented generation [11], a fairness audit in the manner of [3], more capable language models for answer evaluation, integration with LinkedIn, Naukri, Indeed and company applicant tracking systems, deployment on a major cloud platform, Android and iOS applications with notifications, and predictive analytics with personalised learning roadmaps.

References

[1] P. van Esch, J. S. Black, and J. Ferolie, “Marketing AI recruitment: The next phase in job application and selection,” Comput. Hum. Behav., vol. 90, pp. 215–222, 2019, doi: 10.1016/j.chb.2018.09.009. [2] J. Malinowski, T. Keim, O. Wendt, and T. Weitzel, “Matching people and jobs: A bilateral recommendation approach,” in Proc. 39th Hawaii Int. Conf. Syst. Sci. (HICSS), 2006. [3] M. Raghavan, S. Barocas, J. Kleinberg, and K. Levy, “Mitigating bias in algorithmic hiring: Evaluating claims and practices,” in Proc. ACM Conf. Fairness, Accountability, Transparency (FAT*), 2020, doi: 10.1145/3351095.3372828. [4] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “BERT: Pre-training of deep bidirectional transformers for language understanding,” in Proc. NAACL-HLT, 2019, pp. 4171–4186. [5] P. Rajpurkar, J. Zhang, K. Lopyrev, and P. Liang, “SQuAD: 100,000+ questions for machine comprehension of text,” in Proc. Conf. Empirical Methods Natural Lang. Process. (EMNLP), 2016, pp. 2383–2392. [6] N. Reimers and I. Gurevych, “Sentence-BERT: Sentence embeddings using Siamese BERT-networks,” in Proc. EMNLP-IJCNLP, 2019, pp. 3982–3992. [7] Y. Zhang and B. Wallace, “A sensitivity analysis of (and practitioners’ guide to) convolutional neural networks for sentence classification,” in Proc. 8th Int. Joint Conf. Natural Lang. Process. (IJCNLP), 2017, pp. 253–263. [8] T. B. Brown et al., “Language models are few-shot learners,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), vol. 33, 2020, pp. 1877–1901. [9] H. Touvron et al., “LLaMA: Open and efficient foundation language models,” arXiv:2302.13971, 2023. [10] Y. Qin et al., “Tool learning with foundation models,” arXiv:2304.08354, 2023. [11] P. Lewis et al., “Retrieval-augmented generation for knowledge-intensive NLP tasks,” in Proc. Adv. Neural Inf. Process. Syst. (NeurIPS), vol. 33, 2020, pp. 9459–9474. [12] T. Wolf et al., “Transformers: State-of-the-art natural language processing,” in Proc. Conf. Empirical Methods Natural Lang. Process.: Syst. Demonstrations (EMNLP), 2020, pp. 38–45. [13] OpenRouter, “OpenRouter API documentation.” [Online]. Available: https://openrouter.ai/docs

Copyright

Copyright © 2026 Dhanalakshmi R, Akshaya S , Chaithanya M, Jahnavi N. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.