Automated Essay Grading

Designed and executed comparative experiments across five machine learning models, including neural networks and ensemble regressors such as XGBoost, to automate essay grading.

Highlights:

• Evaluated models using Quadratic Weighted Kappa (QWK) and other robust metrics
• Implemented feature engineering tailored to textual inputs (TF-IDF, embeddings, hand-crafted features)
• Ensemble strategies and hyperparameter optimization to improve reliability
• Production-ready inference pipeline with batching and monitoring

Tech highlights: PyTorch/Keras, XGBoost, robust cross-validation strategies, and evaluation focused on educational fairness.