CV

PhD

University of California, Riverside

• Data analysis and predictive modeling in high-energy physics using statistical and ML-based techniques.
• Neural architecture search using tensor completion.
• Relevant Coursework.
  • Data Mining Techniques
  • Database Management Systems
  • Probabilistic Models for AI
  • Optimization in Machine Learning
  • Introduction to Deep Learning
  • Advanced Computer Vision.

BS-MS Dual Degree

Indian Institute of Science Education and Research, Kolkata

• Master’s Thesis - Automated Quantum Field theory calculations using SciPy.

Graduate Student Researcher

University of California, Riverside

• Examine Tensor-Train decomposition with EM-algorithm for low-rank tensor completion in efficient hyperparameter search for neural networks.
• Developed search strategies and engineered features for applying the XGBoost decision tree algorithm. Achieved 80% accuracy in separating rare signal events from background noise.
• Constrained model parameters using hypothesis testing and confidence interval estimation.

Kaggle Competitor

kaggle.com

• Forecasted sub-seasonal temperatures for multiple US locations using autoML library PyCaret for tuning CatBoost and TabNet regressors, achieving a top 30% ranking out of 709 teams.
• Deployed a large protein language model(ESM-2) and 3DCNN architecture (ThermoNet) for predicting enzyme variant stability based on melting temperature data, incorporating protein structure analysis with Rosetta and HTMD, placing in the top 40% among 2482 teams.

Team Lead - Data Science Challenge

Lawrence Livermore National Laboratory

• Mentored four undergraduate students in ML techniques, including data preprocessing, hyperparameter tuning, and Neural Networks, tailored to each team member's prior experience.
• Achieved the highest AUC scoreof 0.89 among six teams with a pretrained transformer model for classifying ligand molecules via SMILES strings.
• Trained 3DCNNs based on voxel representation of ligand structures, achieving an accuracy of 71%.

al-folio

• A beautiful, simple, clean, and responsive Jekyll theme for academics.

• TensorFlow Developer Certificate - TensorFlow Certification Program
• IT Automation with Python Professional Certificate - Google

• Deep Learning Specialization - DeepLearning.AI