At Mediaocean, I engineered data pipelines that kept millions of ad records flowing smoothly every single day. Working with PySpark, Databricks, Kafka, and Delta Lake, I designed both batch and real-time workflows that transformed raw ad logs into analytics-ready datasets.One highlight was implementing Change Data Capture (CDC) with Redshift, which made incremental data loads faster and more efficient. I also optimized data lakes with the Medallion Architecture, reducing query times and improving reliability for business-critical reporting

During my internship at Connection Loops, I got my first taste of research based application at the industry level. I built CNN + TCN deep learning model to predict arrhythmia,improved accuracy by training on synthetic data generated via guassian noise.It was here that I strengthened my foundations in Python, SQL, and cloud-based pipelines, and learned the importance of making data both accurate and accessible