I'm a Data Analyst based in the New York City metropolitan area. I enjoy the part of analytics that happens before a dashboard is built: understanding messy source data, checking whether the numbers can be trusted, and turning the results into something a team can use.
Most of my recent work has focused on vehicle telemetry and operational data. I use Python and SQL for data preparation and validation, Tableau for reporting, and AWS for data storage and analysis.
I'm currently completing an M.S. in Mathematics at the City College of New York.
A before-and-after analysis of fleet performance using Geotab telemetry data.
- Processed approximately 1.7 million event-driven records from 12 commercial vehicles
- Built a Python pipeline to create installation-aligned analysis periods
- Added data-quality checks for timestamps, missing signals, cumulative counters, and irregular reporting intervals
- Developed FEI-Lite, a four-component framework covering fuel efficiency, RPM behavior, idling, and driving stability
- Built three Tableau dashboards for fleet- and vehicle-level review
- Identified 4 improving and 8 declining vehicles after installation
Tools: Python, pandas, Geotab API, Tableau, AWS
A physics-first prototype for estimating expected fuel consumption under different operating conditions.
- Built an OLS Physics Baseline using RPM, engine load, and their interaction
- Added a Random Forest model to predict the remaining physics residual
- Improved untouched final-test MAE by 2.64%
- Evaluated the model using chronological and Leave-One-Vehicle-Out validation
- Used SHAP and actual-versus-expected analysis to identify vehicle and operating-condition review signals
The model is presented as a diagnostic prototype rather than a production fuel-efficiency score.
Tools: Python, pandas, scikit-learn, Random Forest, SHAP, Matplotlib
A CAN telemetry and IMU validation project focused on vehicle operating-state analysis.
- Analyzed more than 1.9 million telemetry records
- Built data-validation and feature-engineering workflows for CAN and IMU signals
- Achieved 0.8500 Macro F1 and 0.8333 Balanced Accuracy
- Completed 41 Python-to-Athena validation checks
- Built an interactive Tableau dashboard to communicate the results
Tools: Python, SQL, AWS Athena, CAN/J1939, Tableau
- Data Analysis: Python, pandas, NumPy, SQL, Excel
- Visualization: Tableau, Matplotlib, dashboard design
- Data Engineering: AWS S3, AWS Athena, ETL pipelines, data validation, Geotab API
- Machine Learning: scikit-learn, Random Forest, OLS, XGBoost, SHAP
- Development: Git, GitHub, Jupyter Notebook