Projects

Make air quality monitoring easier by pulling real-time data, analyzing trends, and visualizing pollutant levels with minimal manual work.

Built a Python pipeline to fetch and process NYC air quality data (API). Orchestrated automated updates via Airflow and built an interactive dashboard for monitoring and analysis.

Reduced manual effort by 50%+ and provided a real-time view into pollution patterns, correlations, and threshold exceedances.

Predict stock price movement using historical data and robust time-series modeling.

Performed EDA + feature engineering; implemented and tuned LSTM models to capture temporal dependencies; evaluated using appropriate forecasting metrics.

Achieved strong forecasting performance (reported ~85% accuracy) and improved stability through iterative tuning and data-driven features.

Predict chronic disease risk and communicate results clearly for stakeholders.

Built predictive models with GridSearchCV and compared algorithms (e.g., KNN, Naive Bayes). Built Tableau dashboards to improve explainability.

Improved model accuracy (reported up to ~90%) and increased stakeholder comprehension by ~35% via better visualization.

Combine sentiment signals with real-time trading flows while keeping latency low and deployment repeatable.

Built Python services integrating sentiment models and REST APIs. Fine-tuned BERT on AWS EC2 and deployed with SageMaker. Optimized execution and pipeline flow.

Reduced data latency by ~70% and improved deployment speed (8h → 2h).