M.Sc. Data Science student at SASTRA Deemed University — building intelligent systems, exploring neural architectures, and turning messy data into decisions that matter. Driven by curiosity, grounded in statistics.
I'm a motivated M.Sc. Data Science student at SASTRA Deemed University with a strong foundation in machine learning, Python, and data analytics. I translate real-world problems into data-driven solutions — from computer vision models to exploratory analysis of complex datasets.
I hold government internship experience analysing field-level data and identifying systemic gaps across rural education programmes in Tamil Nadu. Certified across modern AI/ML frameworks including transformers, NLP, and generative AI.
Beyond tech, I'm a Mridangam player who accompanies bhajans, an automotive enthusiast, and an avid traveler — always curious, always learning.
Built a statistically validated ML framework integrating EDA, feature selection, and multiple models (Random Forest, Gradient Boosting, ANN) for diabetes prediction. Gradient Boosting achieved the best stability (76.4% accuracy, ±0.04 SD) with 5-fold cross-validation. Emphasised feature interpretability — Glucose and BMI identified as primary predictors.
Engineered a CNN-based image classification model fine-tuned on ResNet architecture to identify plant diseases from agricultural imagery. Curated and preprocessed a labelled agricultural dataset, applied data augmentation to reduce overfitting, and iterated on hyperparameters to optimise precision, recall, and F1-score across multiple disease categories.
Conducted comprehensive analysis of bank customer data to uncover primary drivers of churn. Through deep EDA and visualisation, identified key correlations between customer demographics, account activity, and attrition rates — translating complex datasets into clear visual stories that inform business decisions.
In-depth EDA on global F1 race data spanning 20+ international circuits. Ranked the top 10 drivers by average points per circuit, visualised performance trends across seasons, and derived insights on driver consistency — highlighting performance variance across street, oval, and mixed-surface tracks.
This study proposes a structured framework integrating statistical analysis with machine learning for diabetes prediction. Unlike existing approaches focused solely on accuracy, this work emphasises statistical feature validation, clinical interpretability, and multi-model comparison — addressing key gaps in healthcare AI decision-making. Gradient Boosting achieved the most stable performance; Glucose and BMI were identified as the strongest predictors through feature importance analysis.
I'm always open to discussing new opportunities, collaborations, research, or a good conversation about AI and data science. Feel free to reach out!