© 2025 JETIR May 2025, Volume 12, Issue 5 www.jetir.

org (ISSN-2349-5162)

AI-BASED MOCK INTERVIEW SIMULATION

SYSTEM FOR JOB PREPARATION \

Dr. Radha Shirbhate Harshal Kulkarni

Department of Artificial Intelligence Department of Artificial Intelligence
and Machine Learning and Machine Learning
G. H. Raisoni College of Engineering G. H. Raisoni College of Engineering
and Management, Pune 412207, and Management, Pune 412207,
Maharashtra, India Maharashtra, India
radha.shirbhate@raisoni.net kulkarniharshal365@gmail.com

Namrata Bidaye Shivani Kangude

Department of Artificial Intelligence Department of Artificial Intelligence
and Machine Learning and Machine Learning
G. H. Raisoni College of Engineering G. H. Raisoni College of Engineering
and Management, Pune 412207, and Management, Pune 412207,
Maharashtra, India Maharashtra, India
namratabidaye33@gmail.com shivanikangude3@gmail.com

ABSTRACT: impactful alternative to conventional mock

In today’s competitive job market, interview interviews.
readiness is essential, yet many candidates lack This research highlights the potential of
access to realistic practice tools. This research integrating generative and multimodal AI in the
presents an AI-powered Mock Interview interview preparation process and opens up
Simulation System designed to offer personalized opportunities for future advancements in
and immersive interview experiences using only behavioural analytics and adaptive learning.
the user's job title, technical stack, and years of
experience as inputs. Based on this data, the system KEYWORDS:
generates relevant, domain-specific interview AI Mock Interview, Sentiment Analysis,
questions using Google Gemini, a cutting-edge Generative AI, Behavioral Feedback, Interview
generative AI model. Simulation.
Developed with modern web technologies like
React, TypeScript, Firebase, and Clerk I. INTRODUCTION:
Authentication, the platform provides a seamless In today’s highly competitive job landscape,
and secure user experience. A key innovation in candidates are expected to perform well not only in
this project is the use of real-time video recording written assessments but also during interviews that
during mock interviews. These recordings are test communication, confidence, and clarity of
processed by OpenAI’s language and vision thought. However, many individuals—especially
models to perform sentiment analysis, offering students and early-career professionals—struggle
detailed feedback on a candidate’s sentiments. This to find platforms that provide meaningful, real-
enables a holistic review of both verbal and non- time interview practice and actionable feedback.
verbal performance—crucial aspects often missed Traditional mock interviews, though helpful, often
in traditional mock interview setups. Preliminary require scheduling with mentors, lack
testing shows that users report improved self- personalization, and may not capture the
awareness, increased confidence, and better behavioural nuances that employers care about.
articulation after using the system. By eliminating To help bridge this gap, this research presents an
the need for resume uploads and incorporating AI-driven Mock Interview Simulation System that
advanced AI feedback mechanisms, this solution offers a realistic and interactive environment for
provides an accessible, cost-effective, and interview preparation. The system uses basic user
JETIR2505888 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org h858
© 2025 JETIR May 2025, Volume 12, Issue 5 www.jetir.org (ISSN-2349-5162)

information such as job role, relevant technologies, both spoken responses and visual cues using
and years of experience to tailor the interview OpenAI’s language and vision models.
session accordingly. Questions are dynamically As a result, the user receives detailed feedback
generated using Google Gemini, a generative AI not just on the correctness or completeness of their
model capable of producing domain-specific and answers, but also on aspects like emotional tone,
context-aware content. This ensures that users and overall presence—factors that play a critical
receive a unique set of questions aligned with their role in real interviews but are often overlooked.
career goals and skill levels. By combining generative AI with sentiment, the
One of the key innovations in this system is the proposed system provides a holistic and accessible
integration of video recording during mock solution to interview preparation. It empowers
interviews. This allows the platform to analyse users to practice confidently and improve not only
what they say, but how they say it.

II. LITERATURE REVIEW:

Table 1: Literature Review

Sr. Ref Paper Title and Publisher Findings

No No.
1. An AI Mock-interview The AI-powered system is designed to
[1] Platform for Interview simulate realistic interviews and evaluate
Performance Analysis, IEEE 2022 candidates remotely. It leverages machine
learning models to provide automated, real-time
feedback and assessments, making the interview
experience both interactive and insightful.
2. A Comprehensive Study and The system aims to address common
[2] Implementation of the Mock challenges faced by candidates during interviews
Interview Simulator with AI and by offering AI-based mock evaluations. It uses
Pose-Based Interaction, IEEE deep learning techniques to analyze responses and
2024 provide constructive suggestions for
improvement.
3. AI-Driven Virtual Mock The system highlights the transformative
[3] Interview Development, IEEE potential of AI in revolutionizing interview
2024 preparation, offering personalized, interactive,
and data-driven mock interviews that are tailored
to each individual’s job role and experience.
4. A Survey of AI-Driven Mock The system uses advanced techniques like
[4] Interviews using GenAIand Retrieval-Augmented Generation (RAG) and
Machine Learning (InterviewX), Quantized Low-Rank Adaptation (QLoRA) to
ICUIS 2024 generate tailored interview questions and deliver
real-time feedback. By combining fine-tuned
language models, live behavioral analysis, and
automated scoring, it provides a scalable and cost-
effective solution for preparing candidates for
interviews.
5. From Practice to Perfection: The system combines Natural Language
[5] AI-Driven Mock Interviews for Processing (NLP), facial expression recognition
Career Success, ICSCNA 2023 using CNN and the FER dataset, and body posture
detection with MediaPipe to offer a well-rounded
evaluation of both a candidate's technical abilities
and soft skills.

III. AIM AND OBJECTIVE:

The aim of this project is to develop an AI- offering personalized, real-time feedback. Unlike
driven mock interview simulation system that traditional tools that focus solely on technical
helps candidates prepare for job interviews by questions, our system takes a more holistic
JETIR2505888 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org h859
© 2025 JETIR May 2025, Volume 12, Issue 5 www.jetir.org (ISSN-2349-5162)

approach—supporting users not just in showcasing based on both technical content and emotional
their knowledge, but also in improving expression. That’s where our system steps in.
communication, confidence, and emotional
expression. V. METHODOLOGY:
The methodology behind our AI-based Mock
A core objective of the system is to customize Interview Simulation System is designed with a
each interview experience based on the user’s job clear goal: to create an intuitive, personalized, and
title, technology stack, and level of experience. deeply insightful interview preparation experience
This ensures that the questions and feedback are for users. Here’s how the system works, step by
relevant and meaningful to each individual. step, from the moment a user lands on the platform
to when they receive detailed feedback:
By combining this text-based sentiment analysis
with domain-specific question generation and 1. User Registration & Authentication
clear, actionable feedback, the system empowers To get started, users sign up or log in using
candidates to improve not only what they say, but secure authentication powered by Clerk. This
how they say it. Ultimately, this project aims to ensures a smooth and safe onboarding process.
provide an accessible, scalable, and cost-effective Whether it's a first-time user or a returning one, the
tool that helps job seekers practice with confidence platform remembers their profile, past interviews,
and walk into real interviews better prepared— and progress—all tied securely to their account.
both technically and emotionally.
2. Interview Setup – Tailored to the User
IV. EXISTING SYSTEM: Once authenticated, users are invited to create a
Today, a variety of platforms exist to help new mock interview session. They’re asked to
candidates prepare for interviews—but many of input three simple but crucial details:
them still fall short in offering a complete and
personalized experience. Most traditional mock  Their job title (e.g., Front-End Developer,
interview tools follow a standard format with pre- Data Scientist)
recorded or generic questions that don’t reflect an
individual’s specific job role or experience level.  The technology stack they specialize in
While some platforms simulate interviews (e.g., React, Python, AWS)
reasonably well, they tend to focus primarily on
technical questions and miss out on the soft skills  Their years of experience
that are just as important—like clear
communication, confidence, and the emotional These inputs help the system tailor each session
tone behind a candidate's response. uniquely to the user’s career path and proficiency
level. No resume uploads are needed—making the
A few tools have started incorporating AI to setup lightweight yet effective.
generate questions based on user input or resumes,
which is a step forward. However, they often lack
meaningful feedback on how the candidate actually
performs during the interview—especially in terms 3. Personalized Question Generation Using
of how they sound, express themselves, or convey Google Gemini
confidence. The core of the system’s intelligence lies in
Even video-based interview platforms typically Google Gemini, a powerful generative AI model.
emphasize recording and reviewing, without going Once the interview is initiated, Gemini
deeper into analyzing the emotional aspects of a dynamically generates interview questions that are
candidate’s answers. These emotional cues—how domain-specific, relevant to the user's job title, and
confidently someone speaks or how engaged they aligned with their experience level.
sound—can make a big difference in a real
interview setting. This means a beginner front-end developer and
a senior data scientist will receive completely
In short, while existing tools help in certain different sets of questions—both equally
areas, there’s still a need for a more well-rounded challenging and appropriate for their roles.
solution—one that combines smart question
generation with deeper, personalized feedback

JETIR2505888 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org h860
© 2025 JETIR May 2025, Volume 12, Issue 5 www.jetir.org (ISSN-2349-5162)

4. Video Recording
To make the mock interview as realistic as 8. Real-Time Scoring System
possible, video recording is now a mandatory part In addition to the detailed feedback, users
of the experience. As soon as the interview starts, receive a scorecard that reflects their overall
the user's webcam is activated and the entire performance in each category. The scoring system
session is recorded. is transparent and updated instantly—helping users
track their growth over time.
5. Speech-to-Text and AI Validation
As the user responds to each question, their 9. Data Storage and User Dashboard
speech is transcribed into text in real-time. This All interview data—questions, answers, videos,
transcription is then analyzed and validated by and feedback—is securely stored in Firebase.
Google Gemini, which compares it to the ideal Users can revisit their dashboard to review past
answer. This step forms the foundation of the interviews, track progress, and prepare better for
technical evaluation. upcoming sessions.

6. Sentiment Analysis from Responses: 10. Accessibility and Scalability

Now comes the advanced, human-like The entire platform is built on modern, cloud-
evaluation. Instead of analyzing facial expressions based technologies, making it accessible from any
or body language directly from video, the platform device with a webcam and internet. There’s no
focuses on the transcribed responses generated need for scheduling, coaching, or expensive
during the mock interview. These transcriptions are subscriptions—this system provides a full-featured
sent to Google Gemini, which analyzes: mock interview lab, right in your browser.

 Tone and confidence: Does the candidate VI. PROPOSED SYSTEM:

sound enthusiastic, uncertain, or assertive? The proposed system is designed to closely
replicate a real-world interview experience by
 Emotional expression: Are the responses integrating advanced AI technologies within an
calm, tense, or well-composed? accessible web platform. It targets students and job
seekers who aim to improve their communication,
 Clarity of speech: Does the language reflect technical knowledge, and overall presentation
organized thinking and professionalism? through realistic mock interview simulations. The
system is structured into three key stages:
This text-based sentiment analysis allows the
system to assess how candidates express
themselves emotionally, offering deep insight into A. Input Stage
their communication style and overall
confidence—crucial elements in real interviews. 1) User Registration
Users begin by signing up using a simple email-
7. Intelligent Feedback Generation based registration process. Authentication is
Once the interview concludes, the platform handled securely using Clerk Authentication,
brings everything together. It generates a ensuring safe access to personalized interview
comprehensive feedback report broken into three sessions. This initial step requires minimal effort
main areas: while providing access to all the platform's
 Technical Accuracy: How well the capabilities.
user answered the questions.
 Communication Skills: Clarity, 2) Selection of Job Profile
structure, and fluency of speech. After registration, users select a job profile that
 Emotional Tone Feedback: Insights aligns with their career goals, such as “Frontend
from sentiment analysis on how confidently Developer” or “Data Analyst.” They also specify
and clearly the user communicated. their technology stack and years of experience.
Each point includes thoughtful, actionable This input allows the platform to generate tailored
suggestions, making the feedback both personal interview questions and assessments specific to the
and practical. This empowers candidates to user’s desired role, ensuring high relevance and
understand not just what they said, but how engagement.
effectively they delivered it enhancing their overall
interview readiness.
JETIR2505888 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org h861
© 2025 JETIR May 2025, Volume 12, Issue 5 www.jetir.org (ISSN-2349-5162)

B. AI-Driven Mock Interview offers insights that help users understand how their
The core of the platform is an AI-powered words may be perceived during real interviews.
interview simulation engine. It uses Google
Gemini, a generative AI model, to dynamically D. Feedback Generation and Scoring
create interview questions relevant to the user’s Based on the sentiment analysis and technical
selected job profile and skill level. The interview evaluation, the platform generates a detailed
process is structured to simulate a real interview feedback report categorized into:
environment with realistic, interactive questioning.
• Technical response accuracy
• Verbal communication effectiveness
• Emotional tone and expressiveness
C. Intelligent Evaluation System
Following the mock interview, the user’s Each feedback point includes clear and
transcribed answers undergo comprehensive actionable suggestions for improvement.
analysis using advanced AI techniques: Additionally, users receive a comprehensive
performance score reflecting their overall
Sentiment Recognition: interview readiness. All session data is securely
The system uses Google Gemini to analyze the stored in Firebase, and users can revisit their
transcribed text of each response, evaluating the personalized dashboard to review past interviews
emotional tone embedded in the language. It and track progress over time.
identifies patterns such as expressions of
confidence, hesitation, or clarity of thought and
VII. SYSTEM ARCHITECTURE:

Fig 1. System Architecture

JETIR2505888 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org h862
© 2025 JETIR May 2025, Volume 12, Issue 5 www.jetir.org (ISSN-2349-5162)

VIII. OUTPUTS: By leveraging Natural Language Processing and

sentiment analysis on transcribed responses, the
system delivers a well-rounded evaluation. It offers
insights into both technical accuracy and the
emotional tone of the candidate’s responses—
capturing elements like confidence, clarity, and
expression. The feedback is clear, actionable, and
immediately available, making it easier for users to
improve with every session.

In essence, this platform bridges the gap between

solo practice and real-world interviews. It equips
Fig 2. Mock Interview
users with the tools and self-awareness they need
to grow, helping them enter actual interviews with
greater confidence, clarity, and readiness.

X. REFERENCES:
[1] Y.-C. Chou, F. R. Wongso, C.-Y. Chao, and H.-Y. Yu,
“An AI Mock-interview Platform for Interview Performance
Analysis,” in Proc. 2022 10th Int. Conf. on Information and
Education Technology (ICIET), Matsue, Japan, 2022, pp. 37–
Fig 3. Overall Feedback
41. doi: 10.1109/ICIET55102.2022.9778999.

[2] B. B. Jadhav, A. Sawant, A. Shah, P. Vemula, A.

Waikar, and S. Yadav, “A Comprehensive Study and
Implementation of the Mock Interview Simulator with AI and
Pose-Based Interaction,” in Proc. 2024 IEEE Int. Conf. on
Computing, Communication and Intelligent Systems
(ICCCIS), 2024. doi: 10.1109/ICCCIS.2024.10530717.

[3] P. K. Mishra, A. K. Arulappan, I.-H. Ra, L. T.

Mariappan, G. G. Rose, and Y.-S. Lee, “AI-Driven Virtual
Mock Interview Development,” in Proc. 2024 Joint 13th Int.
Conf. on Soft Computing and Intelligent Systems and 25th
Fig 4. Detailed Feedback Int. Symp. on Advanced Intelligent Systems (SCIS&ISIS),
2024, pp. 1–4. doi: 10.1109/SCISISIS61014.2024.10760210.

IX. CONCLUSION: [4] J. V. Barpute, O. Wattamwar, S. Pakjade, and S.

This project introduces a modern and intelligent Diwate, “A Survey of AI-Driven Mock Interviews Using
GenAI and Machine Learning (InterviewX),” in Proc. 2024
way to help candidates prepare for job interviews 4th Int. Conf. on Ubiquitous Computing and Intelligent
by combining artificial intelligence with real-time Information Systems (ICUIS), Pune, India, 2024. doi:
feedback. Unlike traditional mock interviews or 10.1109/ICUIS64676.2024.1086631.
expensive coaching sessions, our system offers a
smart, accessible, and personalized solution that [5] R. Umbare, A. Budhodkar, S. Suryawanshi, S.
Udgirkar, P. Aher, and S. Rangdale, “From Practice to
adapts to each user’s role and experience level. It Perfection: AI-Driven Mock Interviews for Career Success,”
generates tailored questions and provides in Proc. Int. Conf. on Sustainable Communication Networks
meaningful feedback that helps users not only and Applications (ICSCNA), Pune, India, 2024, pp. 1250–
focus on what they say but also on how they 1251. doi: 10.1109/ICSCNA63714.2024.10864324.
communicate.

JETIR2505888 Journal of Emerging Technologies and Innovative Research (JETIR) www.jetir.org h863