Data Scientist based in Tehran, Iran — focused on building end-to-end machine learning pipelines that turn raw data into actionable decisions.
I work across the full data science lifecycle: from exploratory analysis and feature engineering through to model training, evaluation, and result communication. My projects span regression, classification, time-series forecasting, and deep learning.
I recently completed a comprehensive Data Science program and have been applying those skills to increasingly complex real-world problems — most recently a multi-model forecasting system for rail ridership with COVID-19 impact analysis.
Languages & Core Libraries
Machine Learning & Deep Learning
Visualization
End-to-end time-series forecasting pipeline predicting daily rail ridership and allocating trains per station across 2019–2022 (including COVID-19 impact).
- Engineered lag, rolling mean, and cyclical calendar features with strict leakage prevention
- Trained and tuned ElasticNet (GridSearchCV + TimeSeriesSplit) and LSTM (per-station sequences, EarlyStopping)
- Ran a 4-method model comparison: global metrics, visual prediction, per-station RMSE, and train allocation accuracy
- ElasticNet outperformed LSTM across all 5 criteria (RMSE: 433 vs 563, R²: 0.48 vs 0.13, station wins: 21/23)
Python scikit-learn TensorFlow/Keras Time Series ElasticNet LSTM
Multi-class classification on the Dry Bean dataset — distinguishing 7 bean varieties from morphological measurements.
- Performed EDA and feature selection
- Compared multiple classifiers with cross-validation
Python scikit-learn Classification
Binary classification model predicting loan approval outcomes.
- Handled class imbalance and missing data
- Evaluated with Precision, Recall, and ROC-AUC
Python scikit-learn Classification
Exploratory analysis of tennis match data — uncovering performance patterns across players, surfaces, and tournaments.
Python Pandas Matplotlib EDA
Structured data analysis project covering cleaning, transformation, visualization, and insight extraction.
Python Pandas Seaborn EDA
- Advanced deep learning architectures (CNN, Transformers for tabular/time-series)
- Model deployment and MLOps basics
- Feature importance and model interpretability (SHAP)