CrackInt

CrackInt is a full-stack AI interview preparation platform built with Next.js and FastAPI, backed by PostgreSQL with modular API routing. I designed and trained a Word2Vec + BiLSTM + CRF NLP pipeline for Named Entity Recognition over résumé and job-posting text, achieving a test micro-F1 of 0.83 on 4,738 annotated résumés and ~0.85 on 6,327 job postings. The platform features an AI tutor chat, targeted interview question practice, skill gap analysis, readiness scoring, and tailored cover letter generation. LLM-enabled features are controlled through configuration flags for reliable deployment flexibility.