AI CAREER PREPARATION ASSISTANT

CareerPrep AI implements a local-first placement-preparation platform that brings resume matching, interview practice, diagnostic assessment, skill profiling, study planning and progress tracking into one workflow. Its principal objective is not to produce more advice but to connect each piece of preparation activity to a measurable, inspectable record. Every score is computed from stored user activity, and every formula can be read and reproduced. The architecture combines React 18 and Vite with Node.js, Express and embedded SQLite. These are established technologies; the contribution lies in their orchestration around a weighted readiness model, a STAR-based speech evaluator, a role-aware planner and a recruiter and peer layer, delivered in a form that can run on a single machine without a paid service. The integration test shows that the central workflow works, and the analytical review shows how the scores respond to student actions and where they saturate. However, the platform’s feedback is heuristic. The readiness score measures activity more than competence, the interview evaluator measures vocabulary and pace and not meaning, and the resume matcher measures textual overlap and not suitability. The system should be used as an assistive tool that helps a student decide what to do next, with responsibility for consequential decisions remaining with the student and with human mentors. The strongest practical features are the personalised loop from profile to plan to practice to updated score, and the openness of the rules. Further evaluation with real students, expert-rated interviews and a larger test suite would turn these promising design properties into evidence.

[1] T. Janz, L. Hellervik and D. C. Gilmore, Behavior Description Interviewing: New, Accurate, Cost Effective. Boston, MA: Allyn and Bacon, 1986. [2] N. Provos and D. Mazières, “A Future-Adaptable Password Scheme,” in Proc. USENIX Annual Technical Conference, FREENIX Track, 1999, pp. 81–91. [3] M. Jones, J. Bradley and N. Sakimura, “JSON Web Token (JWT),” RFC 7519, Internet Engineering Task Force, May 2015. https://www.rfc-editor.org/rfc/rfc7519 [4] R. T. Fielding, Architectural Styles and the Design of Network-based Software Architectures, Ph.D. dissertation, University of California, Irvine, 2000. [5] Web Speech API, W3C Community Group draft specification. https://webaudio.github.io/web-speech-api/ [6] MDN Web Docs, SpeechRecognition interface. Mozilla. https://developer.mozilla.org/en-US/docs/Web/API/SpeechReco gnition [7] OWASP Foundation, Password Storage Cheat Sheet. https://cheatsheetseries.owasp.org/cheatsheets/Password_Storag e_Cheat_Sheet.html [8] SQLite Consortium, SQLite Documentation. https://sqlite.org/docs.html [9] better-sqlite3 contributors, better-sqlite3 library documentation. https://github.com/WiseLibs/better-sqlite3 [10] Meta Open Source, React Documentation. https://react.dev/