Projects

Developed machine learning models on 1000+ high-entropy alloy composition data points to classify phases (FCC, BCC, IM, FCC+BCC) using thermodynamic and atomic descriptors. Achieved 83% accuracy in multi-class prediction and 92–95% in single-class models, identifying key parameters influencing phase stability and alloy design optimization.

Developed a Conditional Generative Adversarial Network (CGAN) to synthesize deformed microstructures of High-Entropy Alloys (HEAs) based on processing conditions like temperature and strain rate. Implemented a contrastive conditional loss to manage sparse experimental data, achieving < 10% mean error in key microstructural features.