The main focus of this project is to develop an intelligent, user-friendly portfolio generation platform and examine the effectiveness of template recommendations using machine learning techniques during the development and deployment phases. To carry out this project, two stages are considered — the frontend/template development phase and the backend/ML integration phase — with the completion of the frontend serving as the baseline milestone. The frontend/template phase covers the design and implementation of interactive UI components, template layouts, and live preview functionality, whereas the backend/ML phase covers API development, data handling, template recommendation algorithms, and export functionalities. The completed system uses data provided by users — including skills, education, experience, and achievements — as the sample input for generating customized portfolios. A recommendation engine, employing embedding-based similarity search and rule-based heuristics, is integrated to suggest templates best suited to the user's industry and expertise. System evaluation considers performance metrics such as recommendation accuracy, export reliability, and user experience responsiveness as control variables. Findings from internal testing indicate that the frontend design and predefined templates significantly improve the user's ability to customize portfolios efficiently during the pre-integration phase. In the post-integration phase, it is observed that the machine learning model enhances personalization, while backend optimizations ensure reliable export. In terms of system-specific factors, it is found that a clean template structure and low-latency API responses improve usability and adoption rates.
Sravallika et al. (Thu,) studied this question.