The YodhaAI project is an AI-driven Defence Examination and SSB (Service Selection Board) Preparation Portal designed to support aspirants preparing for Indian Defence services such as CDS, NDA, AFCAT, and SSB interviews. It provides a personalized, adaptive, and intelligent learning environment through the integration of Machine Learning (ML), Natural Language Processing (NLP), and data visualization. The system employs the MERN (MongoDB, Express.js, React.js, Node.js) architecture for scalable development and incorporates AI models for mock question generation, performance prediction, and personality assessment. YodhaAI bridges the gap between traditional static preparation and data-driven adaptive learning by offering real-time analytics, instant feedback, and individualized improvement recommendations. This paper discusses the system design, implementation methodology, architecture, and expected outcomes of YodhaAI as a unified AI-based Defence preparation assistant.The YodhaAI project presents an Artificial Intelligencedriven platform designed to revolutionize Defence examination and Service Selection Board (SSB) preparation in India. Each year, millions of aspirants prepare for competitive Defence exams such as CDS, NDA, AFCAT, and CAPF, yet face persistent challenges in accessing personalized guidance, adaptive practice, and continuous performance feedback. Traditional coaching methods and online mock platforms fail to provide dynamic analysis or personalized learning paths. YodhaAI addresses these gaps through an intelligent, data-driven, and interactive environment that personalizes preparation for every user. The platform integrates a full-stack MERN architecture (MongoDB, Express.js, React.js, Node.js) to ensure scalability, modularity, and seamless user experience. On top of this infrastructure, AI-powered modules leverage machine learning and Natural Language Processing (NLP) to generate adaptive question sets, assess SSB responses, and provide performance- based recommendations. Models such as TF-IDF, BERT, and Decision Trees are employed for question generation, difficulty prediction, and weak-topic identification, while sentiment and tone analysis are used for personality evaluation tasks. The system provides realtime visualization of user progress using Recharts and D3.js, enabling aspirants to track their learning curve effectively. Additionally, YodhaAI introduces innovative components such as AI-moderated Group Discussion (GD) chatrooms, automated mock interviews, and notification-based updates for upcoming exams and SSB events. The platform aims to act as a digital mentor—offering guidance, analytics, and motivation simultaneously. By merging modern web technologies with intelligent analytics, YodhaAI aspires to enhance exam readiness, self-awareness, and performance efficiency among Defence aspirants, representing a significant step toward transforming Defence education through AI.

The YodhaAI project represents a pioneering step toward transforming Defence examination and SSB preparation through Artificial Intelligence and data-driven personalization. The system bridges a long-standing gap between conventional coaching methods and intelligent digital mentoring by unifying written test simulation, SSB evaluation, and performance analytics within one adaptive platform. Through the integration of machine learning and Natural Language Processing models such as BERT, TF-IDF, Decision Trees, and K-Means clustering, YodhaAI introduces automation in question generation, test evaluation, and personality assessment. The use of MERN architecture ensures scalability, modularity, and efficient data management, while visualization tools like Recharts and D3.js enhance user engagement through interactive dashboards. This synergy between AI intelligence and modern web technology creates a holistic ecosystem that promotes continuous learning and self- improvement among Defence aspirants. The project’s ability to analyze user behavior, detect weak topics, and provide adaptive recommendations empowers candidates to learn strategically rather than traditionally. Moreover, the integration of AI-based SSB response evaluation adds an innovative dimension by assessing psychological and linguistic attributes such as confidence, coherence, and leadership—traits vital for military selection. In conclusion, YodhaAI successfully merges the disciplines of Artificial Intelligence, education, and human psychology to build a comprehensive and adaptive learning system. It not only prepares candidates for Defence examinations but also cultivates the analytical thinking, discipline, and confidence required in the armed forces. As a scalable and extendable platform, future enhancements such as mobile integration, reinforcement learning for adaptive difficulty, and voice-based GD simulations will further elevate YodhaAI into a nextgeneration intelligent assistant for Defence aspirants across India.

[1]Woolf, B. P., Building Intelligent Interactive Tutors, Morgan Kaufmann, 2010. [2]Zhang, L., et al., “Adaptive Assessment with Reinforcement Learning in Personalized Education Systems,” IEEE Trans. Learning Technologies, 2020. [3]Patel, A., Deshmukh, R., “MERN Stack Implementation in Scalable Educational Platforms,” Int. J. of Computer Applications, 2022. [4]Li, H., Wang, J., Zhao, L., “Data Visualization in Learning Analytics,” Int. J. of Educational Technology, 2021. [5]Devlin, J., Chang, M. W., Lee, K., and Toutanova, K., “BERT: Pretraining of Deep Bidirectional Transformers for Language Understanding,” NAACL, 2019. [6]Sahu, P., Gupta, N., “AI-Generated MCQs using Fine-Tuned BERT Models,” Journal of AI Research and Development, 2023. [7]Kumar, R., Mehta, P., “NLP-Driven Evaluation for Defence SSB Assessments,” IEEE Access, 2022. [8]Mehta, K., Shah, V., “AI-Powered Personalized Learning Systems for Defence Exam Preparation,” IEEE Access, 2023. [9]U. Zara, K. Ayub, H. U. Khan, A. Daud, T. Alsahfi and S. Gulzar, “Phishing Website Detection Using Deep Learning Models,” IEEE Access, DOI:10.1109/ACCESS.2024.3486462. [10]Y. Fang, C. Zhang, C. Huang, L. Liu and Y. Yang, “Phishing email detection using improved RCNN model with multi-level vectors and attention mechanism,” IEEE Access, vol. 7, pp. 56329-56340, 2019.. [11]S. M. Al-Rawahi and M. S. Al-Fahdi, “Using machine learning techniques for rising phishing attacks on social networks,” in Proc. IEEE Conf. on Application, Information and Network Security (AINS), Muscat, Oman, 2020, pp. 1-6. [12]Adaptive Learning Pathways: Advancing Personalization in Online Education Platforms — IEEE Conference Publication. [13]An Adaptive Intelligent Tutoring System Powered by Generative AI — IEEE EDUCON 2025 conference. [14]12. Razaque, M. B. H. Frej, D. Sabyrov, A. Shaikhyn, F. Amsaad and A. Oun, “Detection of phishing websites using machine learning,” 2020 IEEE Cloud Summit, pp. 103-107, 2020 [15]Intelligent Tutoring Systems and Adaptive Learning Environments: Teacher Centric Method in AI Augmented Classrooms — Asian Education Learning Review (but deals with IEEE/style systems).