TrainX: The Interviewer Using Generative AI

TrainX: The Interviewer Using Generative AI Assistant Prof. Nidhi Patel (Mentor) S. Madhu Babu (Team Lead) Pranav Wankhade Sarath Reddy Abstract—This paper introduces TrainX, a mock interview platform built using modern web technologies. It is prepared using the IEEEtran L A T EX class to provide a professional format for submission. TrainX allows users to practice interviews in realtime by giving questions from different domains and providing immediate feedback on their answers. The platform is designed to help users improve technical knowledge, problem-solving skills, and communication abilities. This paper also explains the system structure, main features, and tools used to make the platform reliable and easy to use. TrainX aims to support interview preparation and help users gain confidence for real interviews. Index Terms—Generative AI, Mock Interview, AI Chatbot, TrainX, Career Preparation, Gemini AI, React.js, Next.js, Drizzle ORM. I. INTRODUCTION Job interviews are widely recognized as a key and oftentimes stressful component of the recruitment process. Many individuals have trouble properly preparing as a result of a lack of formalized practice, something that may have a negative impact both on confidence and performance. TrainX addresses these challenges with a complete platform of realistic and immersive mock interviews. By simulating real-world interview scenarios, TrainX empowers individuals to practice questions tailored to their individual disciplines and job function while receiving instant, complete feedback designed to hone their answers, communication skills, and general proficiency. TrainX is designed to meet the needs of a variety of users, ranging from newcomers to the job market to professionals looking to hone their interview skills. Because it can run on various platforms, such as the web, Android, and iOS, it can be practiced from anywhere and at any time, thus making interview preparation highly flexible and user-friendly. Longitudinally, user performance is tracked by the system, allowing individuals to identify their areas of strength and weakness and enhance themselves continuously in areas of problem-solving, communication, and professional competency. Apart from individual benefits, there are real gains for educational centers, training centers, and corporations. By making the process of mock interviews streamlined, it saves time and costs generally incurred in planning interview practice sessions. In addition, it ensures that there is equal access to preparation material to all learners, thus creating a level-playing field conducive to practice. It incorporates current frontend and backend technologies, effective questioning and efficient feedback, resulting in a complete and feasible interview preparation strategy. In summary, TrainX aims to simplify, innovate, and improve the preparation for interviews, preparing users to confidently switch over from academic to professional life. II. LITERATURE SURVEY This in-depth analysis outlines the revolutionary impact of Artificial Intelligence (AI) on simulated interview sites, emphasizing developments in adaptive query construction, personalized feedback, and data-driven career counseling. As the employment market becomes increasingly competitive, AI-empowered interview preparation tools deliver scalable, affordable, and customized preparation choices that enhance candidate engagement and information recall. Several research contributions clarify these innovations and related challenges. Sophie Liu’s exploration of the GPT and Gemini API explains how real-time AI-created queries adapt to a candidate’s past performance and thus boost confidence and preparation, aligning with the foundations of transformer-based architectures introduced by Vaswani et al. [1], BERT [2], and GPT models [3], [4]. It, however, also outlines challenges such as high computational costs and limited domain-specific information, suggesting a need for tailoring with specialized datasets to enhance context appropriateness. Rahul Mehta’s work concerning database infrastructure highlights the importance of efficient data storage and data analytics, particularly leveraging PostgreSQL and ORM frameworks such as Drizzle ORM, to track candidate progress and tailor learning paths. He notes potential performance bottlenecks in high-scale systems, encouraging optimization approaches like indexing and caching, similar to scalable frameworks discussed in enterprise-level AI integration [5]. Liam Wilson advocates the importance of secure measures via role-based access control (RBAC) implemented through Clerk API, and it gives permission to differentiate among candidate, recruiter, and administrator access. Though deployment complexity increases, deployment remains essential for protecting sensitive user information, as highlighted by model transparency practices such as Model Cards for Model Reporting [6]. Maria Gonzalez examines AIdriven data analytics in career building and shows how careers and performance tracking help users build communication, analytics, and technical skills. She highlights a need for clear data policies and encryption to address privacy and ethical issues, which aligns with the affect-aware and ethical learning systems discussed by D’Mello and Graesser [7]. Existing research from Jane Smith outlines the inadequacy of static, non-adaptive simulated interview sites that lack personalized feedback, highlighting the importance of conversational AI chatbots that adaptively react to user proficiency. Studies by Adamopoulou and Moussiades [8], Feine et al. [9], Zhou et al. [10], and Shum et al. [11] support the use of conversational agents that integrate social cues and empathy to improve user engagement and learning efficiency. Furthermore, works such as Brandtzaeg and Følstad [12] and Skjuve et al. [13] demonstrate how chatbots increase adoption through perceived convenience and anthropomorphism, while Budiu [14] highlights usability considerations in chatbot interfaces. Additionally, Mavropoulos et al. [15] present deep learning-based chatbots for job recommendation systems, supporting the integration of AI in employment assistance, while McCarthy et al. [16] and Li et al. [17] emphasize AI’s expanding role in recruitment and sentiment analysis for candidate evaluation. Drawing on these observations, the TrainX platform incorporates Gemini AI for real-time questioning, React and Next.js for responsive interactivity, Neon PostgreSQL for scalable data stores, and Clerk for secure login. Cross-platform deployment (web, Android, iOS) increases accessibility for a range of users from neophytes to veteran professionals. TrainX also benefits educational and training institutions like career counseling centers, schools, and professional training programs by facilitating automated adaptive mock interviews, cost savings, capacity increases, and standardized candidate assessments. Integration of machine learning enables constant monitoring of user progress and adaptive feedback, linking academic preparation and professional success. In addition, TrainX’s granular metrics provide actionable data to employers and recruiters, enhancing recruitment efficiency and talent quality. By promoting inclusivity and accessibility to individuals from a range of backgrounds (e.g., persons with disabilities or language disadvantages) as explored by Pradhan et al. [18], [19], the platform promotes equitable access to career advancement. In short, AI-driven mock interview sites like TrainX are redefining career preparation, professional readiness, and recruitment protocols. By constant innovation and advanced analytics, they enable candidates to manage complicated recruitment processes, increase confidence, and be job-ready. Such technologies have vast potential to recapitalize the workforce to a highly skilled, qualified, and inclusive global environment. III. PROPOSED METHODOLOGY The envisaged TrainX platform provides real-world job interview simulation in an interactive mode, utilizing Gemini AI to learn continuously from user responses and performance data and thus optimize its questioning patterns and feedback protocols [4]. All interactions, performance data, and progress tracking are stored securely in Neon PostgreSQL, allowing the system to provide in-depth insights into users’ areas of strength and improvement [5]. Role-based, secure authentication via Clerk provides role-specific access and experience to students, professionals, recruiters, and career [18]. In addition to conventional mock interviews, TrainX provides sophisticated features like voice-to-text processing and AI-powered sentiment analysis to evaluate tone, confidence, fluency, and clarity Criteria Existing Platforms Proposed System: TrainX AI Technology Use traditional or limited AI models for question generation Employs Gemini Flash 2.0 API for context-aware question generation Question Flow Static or fixed questions Dynamic and adaptive question flow based on user responses Response Evaluation Manual or basic feedback AI-generated accuracy evaluation for each response Data Storage Minimal or no data tracking Stores user data, session history, and accuracy using Drizzle ORM Authentication Basic or none Secure and personalized sessions via Clerk Authentication Frontend Simple web interface Built using React.js and ShadCN for modern UI Backend Generic API setup Express.js + Node.js integrated with Gemini API Performance Analysis Limited or absent Shows detailed accuracy tables for user performance Accessibility Non-personalized or single user mode Personalized dashboards for each registered user Use Case General interview preparation AI-based mock interviewer offering real-time evaluation TABLE I COMPARISON BETWEEN EXISTING AI INTERVIEW PLATFORMS AND THE PROPOSED TRAINXSYSTEM [17], [20]. Scenario role-plays mimicking decision-making and pressure situation solving under simulated scenarios help users to prepare for both behavioral and technical interview situations [16]. The platform has accessibility as a design consideration, allowing it to handle multi-linguistic support, variant text size, voice-aided directions, and text-to-speech, allowing participation for disabled persons and non-English speaking individuals [19], [21]. By combining Generative AI and an interactive and responsive interface, TrainX provides customized career skill building and education [1], [2]. Educational centers, professional schools, and company recruitment teams may employ the platform to deliver tailored training, monitoring, and skill building guidance [13]. Ongoing tracking of performance coupled with adaptive feedback makes learning data-driven as well as measurable [6]. In contrast to conventional mock interview systems relying upon predefined question banks or static content, TrainX generates questions dynamically based upon industry and role, evaluates responses in real time, and offers individual feedback [8]. It conforms to user skill level variation, professional category, and industry standards and makes interview practice a confidence builder and a performance booster. Developed using React, the AI-powered interface provides smooth, real-time interaction that supports constant skill building, employability, and readiness for employment [10]. By utilizing advanced artificial intelligence, natural language processing, and individualized learning approaches, TrainX redefines traditional interview preparation as a scalable, smart, and immersive experience [11]. Candidates can hone key communication, strategic thinking, and problem-solving skills, bridging the gap between academic preparation and employers’ demands [22]. This strategy empowers individuals to excel in competitive job markets while driving the future of AI-enriched career building. IV. IMPLEMENTATION A. Functional Requirements The functional requirements of the proposed system TrainX are outlined in Table. ID Requirement Description FR1 User Authentication Secure login and session management using Clerk. FR2 Mock Interview Simulation Conduct AI-driven interviews using Gemini Flash 2.0 API. FR3 Accuracy Evaluation Analyze candidate responses and compute accuracy scores. FR4 Scoring System Provide structured result data with percentage accuracy per session. FR5 Real-Time Feedback Display response-wise accuracy and session summary instantly. FR6 Dashboard Visualization Show performance metrics and previous results using ShadCN UI. FR7 Data Management Store user profiles, sessions, and results via Drizzle ORM. FR8 API Integration Enable backend communication through Express.js REST APIs. TABLE II FUNCTIONAL REQUIREMENTS OF TRAINX B. Non-Functional Requirements The non-functional requirements of TrainX are presented in Table. ID Requirement Description NFR1 Performance System must evaluate responses and return accuracy within 3 seconds. NFR2 Scalability Support 1000+ concurrent users without noticeable delay. NFR3 Security Protect user data using Clerk Auth and secure Express.js endpoints. NFR4 Usability Provide a clean, responsive interface built with React.js + ShadCN. NFR5 Reliability Maintain 99.9% uptime through efficient backend design. NFR6 Maintainability Modular architecture using React components, Express routes, and ORM migrations. NFR7 Availability Cloud-hosted system accessible 24/7 across regions. TABLE III NON-FUNCTIONAL REQUIREMENTS OF TRAINX This TrainX platform has been designed utilizing modern web technologies to give a realistic, interactive, and dependable environment to practice interviews. Its final job is to replicate actual interview scenarios and pose questions based on certain areas and give quick, constructive feedback to users to help them continuously enhance their performance and confidence. C. Frontend Implementation TrainX front-end design uses React and Next.js technologies. React allows a dynamic and responsive user interface, making it simpler for users to engage with the platform effortlessly across a range of devices. Next.js uses optimization based on server-side rendering and routing, thereby making it easier and efficient to enable movement throughout the platform. Intuitive user interface design emphasizes both efficiency and usability, allowing new joiners and professionals to gain quick access to interview sessions, respond to questions, and evaluate assessments. Furthermore, the platform incorporates accessibility features to enable equal opportunity for participation among users despite technical skills or devices. D. Backend and Data Management The backend system of TrainX serves a key function in managing user information, interview questions, and responses. Neon PostgreSQL is used to manage user profiles, interview histories, and performance reports securely and efficiently. Neon PostgreSQL, as a relational database, allows questions to be stored in separate levels as per domain and job role, tracks user progress across sessions, and generates reports that highlight areas that may require improvement. For user authentication and security, the portal incorporates Clerk. With the help of Clerk, it is ensured that users are able to create accounts and log in to their profiles safely and update information without compromising personal data. E. Simulation of Interviews and Feedback System TrainX groups interview questions by industry and job function, allowing users to select practice sessions that are most relevant. During an interview session, the system provides questions serially, recording user responses as they are entered. For each response, the site provides instant feedback that highlights strengths, pinpoints areas for improvement, and indicates methods to make answers better. Each feedback mechanism has been designed to be both informative and constructive, allowing users to sit back and reflect on their own performance and learn from mistakes. In the long run, users are able to track their progress along a summary of performance, which provides information about trends in their responses, areas of most frequent error, and improvement in specific skill areas. In addition to facilitating individual users, TrainX offers significant benefits to educational centers, training facilities, and corporations. By automating the process of a mock interview, it reduces the time and monetary costs required for interview practice sessions. User Login / Registration Select Domain / Job Role Start Mock Interview Submit Answer Feedback Generation Progress Tracking End Session / Review Summary Fig. 1. Flow chart of TrainX mock interview process Fig. 2. System Architecture of TrainX Fig. 3. Data Flow Diagram of TrainX V. TECHNOLOGIES USED A. Gemini AI Gemini AI is a large language model (LLM) that serves as the core intelligence of the chatbot. It enables dynamic, real-time generation of interview questions tailored to target industries, evaluates user responses, and provides personalized coaching for both technical and behavioral competencies. The Gemini API facilitates seamless integration between the application and the AI model, allowing efficient question generation and AI-based feedback. Its natural language processing (NLP) capabilities support meaningful interactions, offering contextually relevant guidance to improve communication skills, problem-solving abilities, and overall confidence [4]. B. React React is a widely adopted JavaScript library that powers the front-end interface of the TrainX. Its component-based architecture ensures a responsive, scalable, and interactive user experience. Leveraging React’s virtual DOM, interface updates are fast and efficient, providing real-time feedback during interview simulations. The modularity of React allows the incorporation of features such as voice-to-text processing, animated chat outputs, and interactive, role-based scenarios, making the interview experience engaging and realistic [9]. C. Next.js Next.js is a React-based web application framework responsible for backend logic, API routing, and server-side processing. With support for server-side rendering (SSR) and static site generation (SSG), Next.js enhances the performance, scalability, and loading speed of the chatbot. It manages secure interactions with Gemini AI, processes user inputs, and ensures seamless real-time communication within the platform. D. Neon PostgreSQL Neon PostgreSQL serves as the primary cloud-native relational database for storing user profiles, interview records, AI-generated responses, and performance analytics. The wellstructured database design enables efficient data management across multiple sessions, allowing the system to adapt question difficulty and feedback based on user progress. This progressive learning capability ensures that the chatbot remains context-aware and provides personalized guidance [5]. E. Clerk Authentication Clerk Authentication provides secure user authentication and role-based access control (RBAC), enabling personalized experiences for students, job applicants, corporate recruiters, and career advisors. The system supports multi-factor authentication (MFA), OAuth integrations, and password encryption, ensuring sensitive user data remains protected. Role-based access allows candidates to focus on skill development while mentors and recruiters can monitor, assess, and provide guidance effectively [6]. Fig. 4. TrainX Architecture F. Generative AI (GenAI) Generative AI enhances the intelligence and interactivity of the mock interview platform by dynamically generating industry-specific questions, ensuring each session adapts to the user’s skill level and job role. Users receive personalized, real-time feedback that improves technical knowledge, communication skills, and self-confidence. The GenAI-driven chatbot transforms traditional interview preparation into a highly interactive, adaptive, and effective training experience, simulating realistic hiring practices [3]. G. TrainX Features •The platform combines GenAI with React and Next.js to deliver a smart, conversational chatbot interface. •Users engage in dynamic, interview-style chat sessions, where questions adapt based on previous responses, skill levels, and industry domains. •Neon PostgreSQL tracks conversation history and performance analytics, enabling structured data storage, retrieval, and analysis. •Each session is monitored to provide personalized feedback and adaptive guidance, enhancing user readiness for real interviews. VI. RESULTS AND DISCUSSION Generative AI (GenAI) implemented in the TrainX chatbot has transformed how individuals are preparing for interviews. It makes it smart, flexible, and user-friendly. It differs from old fixed-question or video-mock interviews. TrainX generates new questions, verifies answers immediately, and provides individual feedback tailored to the requirements of a job. It makes users continuously improve their technical, communication, and soft skills for actual interviews. It uses React to build the chatbot, which provides a seamless chat experience like having a real interview. Gemini AI generates questions based on the user’s profession, varies the level of questions as they answer, and provides valuable feedback immediately. It provides step-by-step feedback on areas of strength and areas for improvement, and boosts confidence and competency to continuously perform. Data stored in the system are stored securely in Neon PostgreSQL. This database stores user information, interview history, and scores securely and assists the system to note progress and facilitate continuous learning. Clerk makes certain that every user safely logs in securely based on his or her own level of access, whether it’s students, job applicants, or coaches. One of the best things about TrainX is that it can replicate various types of interviews, including technical, behavioral, and case-based interviews. Users can practice solving problems, manage stress, and inculcate clear thinking under stress. TrainX also evaluates voice, tone, and confidence based on speech-to-text and sentiment analysis. It makes both speaking and body language skills better. It’s designed for everybody. It’s multilingual, text-to-speech, voice commands, and font size to enable individuals with a disability or non-English speaking individuals to access it as effortlessly as possible. They enable equal opportunities for individuals from backgrounds to learn and build skills. In addition, the system has scalability, allowing it to be implemented throughout colleges, training centers, and companies. It is timeand cost-effective since it reduces human trainer dependencies while allowing many users to practice skills simultaneously. Through constant interactions with TrainX, the chatbot increases its level of intelligence as it incorporates lessons from past interactions and sharpens its feedback mechanism, thus ensuring gradual skill upgrading and professional growth. The first version (MVP) already has features like real-time communication, individual feedback, safe access, and AIpowered performance tracking. In the future, it is projected to have sophisticated natural language processing powers, more sophisticated emotional detection, and industry-specific trainable models to boost accuracy and efficiency. Overall, this AI-based chatbot makes interview preparation efficient, making it both simple and efficient. It enables individuals to gain confidence, sharpen their capacity to solve problems, and grasp industry standards. In the long run, it intends to deliver skilled, job-ready candidates. Fig. 5. TrainX - Home page Fig. 6. TrainX - Dashboard Fig. 7. TrainX - Dashboard (Extended) Fig. 8. TrainX - Interview (Part 1) Fig. 9. TrainX - Interview (Part 2) Fig. 10. TrainX - Feedback Page VII. CONCLUSION Combining Generative AI (GenAI) and AI-powered mock interview simulations represents a paradigm shift in the landscape of careers preparation and job readiness. This system not only enhances candidate conduct but also increases confidence and flexibility in real-world interview situations. By combining chat AI functionality with real-time adaptive learning modalities, the platform corrects knowledge gaps, tailors practice modules based on user demand, and offers highly interactive and structured interview preparation regimens. With context-aware interactions, the chatbot dynamically generates relevant and industry-specific questions while, in parallel, evaluating responses to provide real-time feedback about technical, behavioral, and communication skills. Acting as a virtual career coach, the system walks users step-bystep, delivering formal and constructive evaluations designed to hone their interviewing skills. With React-created front-end, the system provides a seamless and engaging user experience, and Gemini AI’s sophisticated natural language processing (NLP) capabilities enable intelligent query construction and adaptive conversational flows. Additional modules, such as voice-to-text, sentiment analysis, and scenario-based simulations, enhance the practice environment, allowing candidates to enhance verbal and non-verbal communication skills in real time. Neon PostgreSQL provides the platform data storage base, delivering scalable and secure data storage options for user histories, metrics, and progress tracking. Role-dependent data encryption, managed by Clerk, ensures privacy and tailored access for various user groups—students, searchers, professionals, recruiters, and career consultants. This architecture provides secure yet role-dependent functioning across the system. Scalability and flexibility of the platform make it suitable for deployment in educational institutions, training academies, and workplace recruitments. By automation and standardization of mock interviews, organizations can leverage streamlined preparation procedures while ensuring costeffectiveness along with consistency. Aside from supporting individual professional growth, the system uses user activity and performance data to generate key insights, thus building a constant feedback loop that optimizes and upgrades the AI models over time. This constant improvement ensures that the chatbot gets progressively more efficient, contextually astute, and consistent with the evolution of the industry standards. Behind every vital aspect, the essential value of the project lies in the fact that it fills the gap between academic education and real-world employers’ expectations, thus presenting candidates with wiser, scalable, and tailored training methods that greatly boost career readiness. By instilling confidence, skill-building, and flexibility, the AI-powered mock interview platform not just gets individuals ready to nail successful interviews, but largely transforms the broader recruitment and professional growth context. VIII. FUTURE SCOPE The TrainX system has enormous scope for growth and enhancement in the future, aimed towards making preparation for interviews efficient, accessible, and customized for users around the world. One key direction for growth involves the addition of domain-specific mock interviews covering a vast range of professional fields, including software engineering, finance, healthcare, marketing, and many others. With this, users will have the opportunity to practice questions that are directly relevant to their professional paths, thereby increasing the relevance and quality of preparation. In addition, a meaningful upgrade involves the addition of more advanced feedback mechanisms that not just assess the correctness of answers but also user behavior, including levels of stress, confidence, and communication skills throughout the interview process. By incorporating these feedbacks, users will gain complete feedback that goes beyond correctness, allowing them to identify areas that require improvement and develop better professional skills in the long run. TrainX is further aimed at supporting user learning and career growth in the long run, introduced the concept of AI-powered career dashboards. The dashboards will provide individualized recommendations for skill upgrade, suggest places to learn, and track progress in the long run, thus allowing users to focus on their weaker areas in a systematic approach. In addition, the platform will be extended to include multilingual support and cultural flexibility, enabling individuals from various nations and linguistic backgrounds to avail the system. This flexibility ensures that TrainX remains accessible and functional to a global user base. Future studies will also focus highly on fairness and inclusivity in assessment procedures. Measures will be introduced that address possible biases in computerized feedback systems, ensuring that every user, regardless of their background or experience, gets equitable and uniform assessments. Also, the platform will have peer and collaborative learning capabilities, enabling learners to converse, practice, and exchange ideas with other learners to make the learning process more engaging. Finally, TrainX will further upgrade its technology platform to make it scale, secure, and perform repeatedly across a variety of devices. Keeping these developments in focus, TrainX aspires to become a complete, intelligent, and user-obsessed platform for professional development, empowering individuals to gain confidence, acquire necessary skills, and thrive in their careers globally. REFERENCES [1] A. Vaswani, N. Shazeer, N. Parmar et al., “Attention is all you need,” in NeurIPS, 2017. [2] J. Devlin, M.-W. Chang, K. Lee, and K. Toutanova, “Bert: Pre-training of deep bidirectional transformers for language understanding,” in NAACL, 2019. [3] A. Radford, J. Wu, R. Child et al., “Language models for unsupervised text generation,” OpenAI, Tech. Rep., 2019. [4] T. Brown, B. Mann, N. Ryder et al., “Language models are fewshot learners,” in Advances in Neural Information Processing Systems (NeurIPS), 2020. [5] A. Diederich, S. Brendel, and T. Korn, “Ai in recruitment: A review of ai applications in human resource management,” Applied Sciences, vol. 12, no. 5, 2022. [6] J. Clark and M. Mitchell, “Model cards for model reporting,” in FAT Conference, 2019. [7] S. D’Mello and A. Graesser, “Feeling, thinking, and computing with affect-aware learning technologies,” International Journal of Artificial Intelligence in Education, vol. 25, no. 2, pp. 144–158, 2015. [8] E. Adamopoulou and L. Moussiades, “An overview of chatbot technology,” Artificial Intelligence Applications and Innovations, vol. 20, no. 4, pp. 1–12, 2020. [9] J. Feine, U. Gnewuch, S. Morana, and A. Maedche, “A taxonomy of social cues for conversational agents,” International Journal of HumanComputer Studies, vol. 132, pp. 138–161, 2019. [10] L. Zhou, J. Gao, D. Li, and H.-Y. Shum, “The design and implementation of xiaoice, an empathetic social chatbot,” Computational Linguistics, vol. 46, no. 1, pp. 53–93, 2020. [11] H.-Y. Shum, X. He, and D. Li, “From eliza to xiaoice: Challenges and opportunities with social chatbots,” Frontiers of Information Technology & Electronic Engineering, vol. 19, no. 1, pp. 10–26, 2018. [12] P. B. Brandtzaeg and A. Følstad, “Chatbots: Changing user needs and motivations,” Interacting with Computers, vol. 30, no. 3, pp. 263–275, 2018. [13] M. Skjuve, A. Følstad, and P. B. Brandtzaeg, “Adoption of chatbots for customer service: The roles of perceived convenience and anthropomorphism,” Journal of Business Research, vol. 125, pp. 381–390, 2021. [14] R. Budiu, “The ux of chatbots,” Nielsen Norman Group, 2018, online Article. [15] A. Mavropoulos, K. Ismini, and M. Iosif, “A deep learning-based chatbot for job recommendation systems,” in Conference on AI & NLP, 2020. [16] M. McCarthy, K. Hughes, and A. Hill, “The role of artificial intelligence in enhancing job interview processes,” Journal of Business Research, vol. 145, pp. 120–132, 2022. [17] H. Li, Y. Liu, and J. Wang, “Sentiment analysis for employment screening: A survey,” IEEE Access, vol. 10, pp. 50 234–50 250, 2022. [18] B. Pradhan, R. Mehta, and L. Findlater, “Using chatbots to support individuals with visual impairments in social media platforms,” in Proceedings of the 21st International ACM SIGACCESS Conference on Computers and Accessibility (ASSETS ’19), 2019. [19] A. Pradhan, A. Lazar, L. Findlater, and J. Froehlich, “Use of chatbots in disability support: A case study in visual impairment assistance,” ACM Transactions on Accessible Computing (TACCESS), vol. 13, no. 4, p. 19, 2020. [20] N. Novielli, F. Calefato, and F. Lanubile, “A systematic review of sentiment analysis in software engineering,” Information and Software Technology, vol. 58, pp. 75–94, 2015. [21] D. P´ erez-Marcos, “Virtual reality and artificial intelligence in patient education and behavioral health interventions,” Journal of Medical Internet Research, vol. 23, no. 6, p. e25312, 2021. [22] P. Schmitt, J. Bechter, and J. Schulze, “Chatbot adoption in business: Why users are willing to use chatbots,” Journal of Service Management Research, vol. 5, no. 1, pp. 41–53, 2021.