Visit my personal website ➡️➡️ wardiyahrammazy.com

I also casually post articles about tech and math on my newsletter Dev-o-Ramma which you can find at devoramma.substack.com

• Did my first ever datathon! The WiDS Worldwide Datathon 2026 with predictive modeling for Climate/Fire Impact.

Most Used Programming Languages/Frameworks: JavaScript, TypeScript, HTML, CSS, Python (Data Science Focus), R, SQL, React, Expo

Specialized Data Libraries: Scikit-Survival, XGBoost, LightGBM, CatBoost, Optuna, TensorFlow, Scikit-learn, Pandas, NumPy

Tools & Technologies: Google Colab, Jupyter Notebook, Looker Studio, Git, Figma, Visual Studio, VSCode

Technical Expertise: Machine Learning, Survival Analysis, Feature Engineering (Physics-Informed), Web Development (Frontend-Heavy Fullstack)

Predicting the probability of wildfire impact across 12h, 24h, 48h, and 72h horizons using survival analysis and ensemble learning.

• Physics-Informed Engineering: Created domain-specific features like Wavefront ETA (combining radial growth and movement) and Near-Miss Margin to capture fire dynamics.
• The Blend: Implemented a hybrid ensemble of Random Survival Forests (RSF) and Gradient Boosting (XGB/LGB/Cat), achieving a localized Brier Score of 0.003.
• Statistical Rigor: Managed right-censored data and ensured logical consistency through monotonicity constraints and 5-Fold Stratified Cross-Validation.

Languages & Libraries: Python, Scikit-Survival, XGBoost, Optuna, Matplotlib, Seaborn

Developed as part of the Break Through Tech AI Program, this project automates resource matching for low-income social entrepreneurs.

• Problem: Small business owners waste significant time manually searching for localized growth resources.
• Solution: Built a scalable recommendation system using Random Forest and KNN (most accurate) to match demographic data to high-impact resources.
• Data & Features: Engineered features from technology access and income levels; used NLP on open-ended responses for deeper categorization.

Languages & Libraries: Python, TensorFlow, Scikit-learn, Matplotlib

A collaborative neighborhood safety application.

• Lead Role: Served as Frontend Lead, designing the UI/UX in Figma and implementing the architecture in Expo/React Native.
• Fullstack Contributions: Managed user authentication, email verification, and backend integration for account management.

Languages & Frameworks: TypeScript, React Native, Expo, Node.js