http://hdl.handle.net/123456789/10320 Sat, 04 Apr 2026 10:48:23 GMT 2026-04-04T10:48:23Z http://hdl.handle.net/123456789/19631 KuickBoard - AI Driven Employee On boarding and Assessment Platform Muhammad Fawaz Khalid, 01-131212-021; Muhammad Gulsher Khan, 01-131212-022 Integrating new employees into an organization, commonly known as onboarding, is a very critical process in any company but is often implemented ineffectively. Traditional methods rely on methods that are manual, time-consuming, costly and may also fail to provide any customization to the training on onboarding process. This project “KuickBoard” addresses these challenges trough a platform made to streamline and automate the key functions of employee assessments and onboarding by generating courses and assessments using Artificial Intelligence. The purpose behind KuickBoard is to lessen the burden on the administration, make training customizable and easy, and encourage employee engagement and skill development from the first day that they enter the company. KuickBoard was implemented as a web application. The frontend user interface was made using Angular because of its component-based structure. The backend of the website was built using .NET Core framework. The system architecture follows the Model-View-ViewModel (MVVM) pattern, promoting separation of concerns. Data storage is handled by a MySQL relational database, and user authentication is managed via JSON Web Tokens (JWT). The AI used for course and assessment generation were mainly ChatGPT 3.5 turbo and Gemini 2.0 Flash. The KuickBoard prototype has a multitude of core functionalities. These include user management for both individual learners and company employees, capabilities for companies to manage their profiles, teams, roles, and employee invitations, and most importantly, AI-powered features for generating candidate assessments and employee training courses based on prompts. The platform also includes mechanisms for assigning courses to employees and tracking their progress. In conclusion, this project is a platform that demonstrates the potential benefits of an AI-powered approach to employee onboarding and assessment. It successfully meets the primary goals of automating training and providing a centralized system. The platform is a step towards more efficient, scalable, and potentially customizable corporate onboarding strategies. While acknowledging the necessity for future work, particularly in areas such as AI bias prevention, advanced analytics, broader third-party system integrations, and comprehensive user testing, KuickBoard effectively lays the groundwork for a next-generation onboarding solution. Supervised by Dr. Adeel Muzaffar Sayed Wed, 01 Jan 2025 00:00:00 GMT http://hdl.handle.net/123456789/19631 2025-01-01T00:00:00Z http://hdl.handle.net/123456789/19625 Smart Prep AI Ayesha Maryam, 01-131212-009; Zarina Attaria, 01-131212-040 SmartPrep AI is an intelligent, cross-platform web-based application designed to revolutionize the way job seekers prepare for interviews. With the rising demand for employment and the competitive nature of job markets, interview readiness has become essential. Traditional mock interview methods often lack accessibility, personalization, and real-time feedback. To address these limitations, SmartPrep AI provides a complete AI-driven interview simulation platform that helps users enhance their verbal and non-verbal communication skills. The main motivation behind developing this system was to create an affordable, accessible, and intelligent platform that uses modern technologies to bridge the gap between academic knowledge and professional readiness. In SmartPrep AI, users can register as job seekers and take part in text-based or video-based interview simulations. The system generates questions relevant to the user’s selected domain and analyzes responses, Speech Analysis, and Body Language Detection. Real-time AI feedback is provided on various parameters such as tone, content relevance, knowledge, posture, emotion, speech analysis, eye contact, and hand gestures. A progress report system tracks user improvements over time, while an integrated job portal allows users to search and apply for jobs. Our application is developed using the MERN stack (MongoDB, Express.js, React.js, Node.js) for frontend and backend development, while Python-based AI modules are used for video and audio analysis. We integrated external AI APIs for response evaluation and feedback generation. For documentation and modeling purposes, Visual Paradigm was used for class, use case, and ER diagrams, and MS Word was used for report writing. Chapter 1 of the SmartPrep AI report introduces the project, its motivation, and methodology. Chapter 2 presents the background and literature review highlighting the need for AI-based interview training. Chapter 3 defines system requirements and includes detailed use case diagrams. Chapter 4 covers system design with architecture, flow, and data diagrams. Chapter 5 elaborates on the implementation with screenshots and technical details. Chapter 6 explains system testing through detailed test cases, and Chapter 7 concludes the report with the project’s impact, challenges, and future enhancements. Supervised by Engr. Rafia Hassan Wed, 01 Jan 2025 00:00:00 GMT http://hdl.handle.net/123456789/19625 2025-01-01T00:00:00Z http://hdl.handle.net/123456789/19632 Dream– Dyslexia Recognition and Early Assessment Module Minahil Naseer, 01-131212-018; Hamna Kashif, 01-131212-052 The early detection of learning disabilities such as dyslexia, dyscalculia and dysgraphia are crucial for providing timely intervention and support to children at a very young age. Identifying these disabilities at an early stage can significantly improve a child’s academic performance and overall development. However, traditional methods of detection often require specialized expertise, time, and resources, making them inaccessible to many parents and educators. To address this challenge, we developed a framework named as Dyslexia Recognition and Early Assessment Module (DREAM) to streamline the detection process through interactive and technology-driven assessment modules. DREAM provides a secure login and registration system, ensuring that only parents and teachers can access the platform. Upon reaching the main menu there are three distinct detection modules: Dyscalculia Detection, Dysgraphia Detection and Dyslexia Detection. The Dyscalculia Detection module employs a quiz-style assessment powered by Machine learning (ML) algorithms to analyse the likelihood of difficulties in the child’s mathematical skills or abilities. The Dysgraphia Detection module leverages Convolutional Neural Network (CNN) of Deep Learning (DL) to assess the child’s handwriting based on their writing styles and word formation. The Dyslexia Detection module is an interactive game that comprises of four levels – Sight Words, Audio Detection, Colour and Letter Sequence and Phonics Reading – each are tailored to evaluate specific cognitive and linguistic skills. In addition to detection modules, DREAM provides profile section where parents or teachers can view detailed reports of their child’s performance along with suggestions on how they can improve their child’s performance, modify details and access help resources. DREAM is built using Flutter, Python and Firebase, enabling efficient data processing, report generation, and secure storage of user information. By integrating advanced machine learning and deep learning techniques, it provides accurate and insightful assessments, making it a valuable tool for parents and educators. Supervised by Dr. Kashif Sultan Wed, 01 Jan 2025 00:00:00 GMT http://hdl.handle.net/123456789/19632 2025-01-01T00:00:00Z http://hdl.handle.net/123456789/19627 Digital Stethoscope with Companion Mobile App (AIRIS) Laiba Nasir, 01-131212-017; Ali Abbas Kazmi, 01-131212-046 Respiratory illness remains a significant cause of morbidity and mortality throughout the world, particularly in low-resource environments where direct access to diagnostic tools and healthcare professionals is not always feasible. Traditional diagnosis by auscultation and pulmonary function tests usually relies on the presence of specialized equipment and trained healthcare professionals, so early diagnosis and continuous monitoring become unattainable for most. Such absence of readily accessible respiratory care significantly prolongs diagnosis and treatment and leads to more serious health consequences. To overcome this challenge, we introduce AIRIS (Advanced Insight for Respiratory Intelligence System)—a software and hardware combined solution whose aim is to support early detection and monitoring of respiratory diseases. The system is equipped with a digital stethoscope specially tailored for the recording of high-fidelity lung sounds, driven by a Raspberry Pi Zero 2W. The audio signals are sent to a mobile app, built using Flutter, where real-time noise reduction, visualization, and analysis are performed. At its heart is a machine learning model based on Convolutional Neural Networks (CNN), trained on the ICBHI dataset for respiratory sound classification and detection of potential anomalies. The AIRIS app also supports an end-to-end healthcare process with secure login by users, doctor-patient messaging, appointment scheduling, and patient monitoring of progress—backed by Firebase. There is also an AI-driven chatbot with LangChain and OpenAI APIs to support users in navigating symptom-related questions. With smart diagnostics, remote access, and ease of use, AIRIS is an affordable and scalable solution to respiratory healthcare, particularly valuable for underserved and remote populations. Supervised by Dr. Kashif Sultan Wed, 01 Jan 2025 00:00:00 GMT http://hdl.handle.net/123456789/19627 2025-01-01T00:00:00Z