• Crafted a full-stack job market intelligence platform in Python and FastAPI to replace unreliable single-source job data collection, exposing 8 REST API endpoints across 2 job data sources. • Implemented a 3-layer skill-extraction pipeline combining regex pattern matching and machine learning-based semantic matching, and trained a role-classification model using Scikit-learn to automatically categorize job postings. • Developed a resume-matching feature using FAISS and Sentence Transformers for semantic search, enabling automated resume-to-job skill gap analysis across 500+ job postings.

• Resolved franchise renaming inconsistencies that broke aggregate SQL queries by building a Python ETL pipeline to normalize 295,732 records across 1,243 matches into a structured SQL data warehouse. • Addressed data leakage risk from naive train-test splits by training a LightGBM machine learning model using season-based holdout validation, achieving 78% accuracy and a 0.90 ROC-AUC score. • Replaced static SQL-only analytics with a full-stack solution by developing a FastAPI backend and React dashboard, enabling real-time win-probability predictions through REST API endpoints.