Sushma Bandugula

About Sushma Bandugula

I’ve always been captivated by the power of data and its ability to tell stories that drive decisions. My journey started at Novartis India where I built robust data pipelines for clinical trial data. This experience not only sharpened my technical skills but also ignited my passion for transforming complex datasets into actionable insights. As I navigate the ever-evolving landscape of data engineering and science, I find joy in uncovering patterns that can lead to impactful business decisions.At Rocktop Technologies, I developed predictive models using Python and Scikit-learn to analyze over 15 million daily transactions. This effort enhanced fraud detection accuracy by an impressive 32%. Prior to this role, my time at Capgemini was equally rewarding; I created machine learning models for customer segmentation that improved client retention rates by 18% while enhancing demand forecasting accuracy by 22%. It’s fulfilling to see how data-driven decisions make a tangible difference.My background also includes working at Tata Consultancy Services where I engineered scalable ETL pipelines using Apache Airflow and Python to process over 10 million healthcare records monthly. This significantly enhanced patient data management efficiency across the organization. Each of these experiences has contributed to my understanding of how vital accurate data analysis is in today’s world.When I\'m not knee-deep in code or analyzing datasets, you might find me attending local tech meetups or engaging with communities focused on harnessing the power of analytics for social good. I believe that sharing knowledge and collaborating with others is just as important as personal development. Let’s connect and talk data! Let’s connect or reach out at: s••••••••@gmail.com

Experience

1. Data Scientist

   Rocktop Technologies

   Oct 2024 — Present · Houston, TX, US

   Developed predictive models using Python and Scikit-learn, enhancing fraud detection accuracy by 32% while simultaneously reducing alert noise by 25%, resulting in more efficient operations.Created automated machine learning pipelines with Apache Airflow that cut model retraining time by 40%, enabling near real-time risk scoring for global trading teams and boosting decision-making agility.Designed Snowflake-powered analytical datasets that transformed critical risk assessments, decreasing reporting time from two hours to just ten minutes, significantly improving operational efficiency.Integrated Kafka streaming feeds for sub-second anomaly detection across business units, enhancing operational transparency while ensuring compliance with regulatory standards through rigorous data validation protocols.

Education

• Lindsey Wilson University

  Master's Degree, Management Science