SARATH AC

Dynamic AI/ML Intern at CDAC, adept at enhancing neural network performance through optimized machine learning algorithms. Proficient in Python and AWS, I successfully led project proposals and presentations, showcasing strong communication skills and a commitment to innovation in data-driven solutions.

End-to-End ML Project: Student Exam Score Prediction

• Developed an end-to-end ML pipeline in Python for predicting student exam scores.
• Implemented data ingestion and preprocessing using Scikit-learn (ColumnTransformer).
• Trained models with XGBoost, CatBoost, and optimized hyperparameters using GridSearchCV, achieving ~0.87 R² score.
• Built a modular prediction pipeline with exception handling and logging for reliable inference.
• Deployed the solution on AWS Elastic Beanstalk using a CI/CD pipeline, ensuring automated and scalable delivery.

End-to-End Bank Loan Eligibility Analysis

• Analyzed financial data of 3,000 customers to identify factors linked to loan default.
• Performed data cleaning (handling missing values, removing duplicates, encoding categorical variables, outlier detection) using Pandas & NumPy.
• Conducted feature engineering (binning income levels, debt-to-income ratio, credit score grouping) to improve prediction accuracy.
• Performed Exploratory Data Analysis (EDA) with Matplotlib & Seaborn to visualize patterns in approval rates, defaults, and customer profiles.
• Designed an interactive Power BI dashboard for stakeholders to monitor approval trends, high-risk segments, and default likelihood.

Google Data Analyst professional certificate (Coursera)