Pritesh P.

Led the collection, cleaning, and formatting of raw datasets utilizing Pandas and NumPy to support automated workflows and scalable ETL pipelines. Conducted exploratory analysis and statistical testing to comprehend feature distributions, identify outliers, and analyze correlations affecting algorithm performance. Developed predictive features to enhance model accuracy and improve dataset usability for enterprise business intelligence reporting. Trained algorithms, including logistic regression, decision trees, random forests, and k-nearest neighbors through Scikit-Learn. Evaluated system performance with industry-standard metrics such as precision, recall, F1 score, ROC AUC, RMSE, and R-squared. Created regression pipelines to predict laptop pricing based on hardware specifications and brand attributes. Designed a convolutional neural network (CNN) using TensorFlow and Keras to classify brain tumor images from MRI scans. Implemented image preprocessing and augmentation techniques, such as resizing, normalization, and rotation, to enhance network generalization. Developed a collaborative filtering engine to generate personalized book recommendations based on user-item interaction metrics. Employed k-nearest neighbors (KNN) similarity techniques to discern relevant items through behavioral pattern analysis. Tuned algorithms via hyperparameter optimization methods like grid search and cross-validation to ensure stability in production environments. Crafted interactive dashboards with Matplotlib and Seaborn to deliver clear insights to stakeholders. Wrote complex SQL queries for data extraction and analysis from relational database management systems. Utilized advanced gradient boosting frameworks, including XGBoost, to improve predictive performance on complex datasets. Experimented with deep learning architectures in PyTorch to address advanced classification and pattern recognition tasks. Built Natural Language Processing (NLP) pipelines with Hugging Face Transformers to extract actionable insights from unstructured text data.