Mehak

Built a full-stack AI-powered expense management application using the PERN stack with JWT authentication, enabling secure management of transactions, budgets, and financial categories.

• Designed responsive dashboards with real-time financial analytics, spending trends, and budget tracking using React, Tailwind CSS, and Recharts for improved user insights.
• Integrated Google Gemini AI to deliver personalized financial summaries and actionable savings recommendations by analyzing users' spending patterns.

Engineered a backend news pipeline processing 50+ articles per run using TextRank and BART-based summarization.

• Built a context-aware Q&A module with transformer models, supporting multiple queries per article.
• Added error handling and fallback logic, achieving >99% successful API ingestion.
• Structured modular, documented code aligned with production standards.

Analyzed and transformed 1,000+ customer records through data cleaning, encoding, and feature engineering.

• Trained and benchmarked 3+ supervised classifiers, delivering 82%+ test accuracy in churn prediction.
• Assessed model performance using precision, recall, and ROC-AUC to enable data-driven retention insights.