Hey! I am Sahithya 👋

Developed time series forecasting models, including ETS, Holt-Winters, and SARIMA, to predict oil sales trends by analyzing historical data, imputing missing values, and identifying seasonal patterns for actionable insights.

Developed a distributed machine learning pipeline using PySpark MLlib to preprocess a large dataset of 63,003 records and optimize models like Logistic Regression and Gradient Boosting for diabetes risk prediction.

Developed a comprehensive database system for an online shopping platform by programmatically generating synthetic data tables using Python’s Faker module; created datasets for testing database queries, indexing, and performance analysis.

Implemented a neural network from scratch in Python for MNIST digit classification, including preprocessing, backpropagation, and optimization, achieving accurate predictions through custom training and evaluation.

Implemented LDA, QDA, and Ridge Regression from scratch in Python, optimizing classification and regression tasks with feature engineering, regularization, and polynomial mapping for high accuracy.

Implemented logistic regression and support vector machines (SVMs) for MNIST digit classification using Python, achieving high accuracy through gradient optimization, multi-class extensions, and hyperparameter tuning for SVMs.

Built a property price prediction system using data normalization, preprocessing, model training, and optimization; selected the best model, deployed it via a Docker image on Digital Ocean, and developed a Streamlit app for user interaction.

Developed a predictive model to classify obesity risk groups using machine learning, including SVM, KNN, Random Forest, and Decision Tree, with preprocessing, EDA, and performance evaluation (accuracy, precision, recall, F1 score)

Created a dashboard to analyze the impact of bank marketing campaigns on term deposit subscriptions, featuring forecasts, bar graphs, tables, stacked charts, and filters for detailed insights.

Built a Retrieval-Augmented Generation (RAG) application in Python to enable querying and chatting with PDFs using generative AI, covering local deployment, vector database updates, file handling, and response quality testing.