my.projects
This is a collection of my projects, where I've focused on extracting insights and developing strategies. I've worked to make the findings clear and accessible. Take a look and see what you think.
UT DALLAS CHAT BOT
- Developed a RAG pipeline using sentence-BERT for UT Dallas website knowledge base, achieving 92% top-1 retrieval accuracy
- Implemented a GDPR-compliant user query anonymization module using regex-based PII detection with less than 2% false positive rate.
- Alternatively proposed Zapier-based architecture reducing costs while enabling 5x faster deployment through no-code automation
PythonRAGsentence-BERTZapierGDPR
View ProjectCREDIT RISK MODELING
- Built a credit risk classification model using XGBoost-based feature selection and neural networks on the American Express default prediction dataset, identifying high-risk applicants through probabilistic scoring
- Demonstrated a 0.9% improvement in default prediction accuracy over baseline model, enabling potential for a 2% lift in simulated portfolio profitability through selective approval thresholds and risk-based segmentation
PythonXGBoostNeural NetworksMachine Learning
View ProjectCONAGRA BRAND STRATEGY
- Analyzed consumer behavior patterns and market trends to develop comprehensive brand positioning strategies
- Created data-driven recommendations for product portfolio optimization and market expansion
- Delivered actionable insights that informed strategic decision-making for brand growth initiatives
Market AnalysisStrategyData VisualizationBusiness Intelligence
View ProjectEV CHARGING OPTIMIZATION
- Increased charging station service rates by 17.2% and revenue by 6.3% through dynamic pricing tiers
- Scaled solutions to 150+ stations across 3 municipal EV programs, reducing peak grid load by 22%
OptimizationEnergy SystemsSmart GridSustainability
View Project