Niranjini Kathiravan

Master Thesis: Remaining Useful Life Prediction of Hydraulic Pumps (Sep 2024 – Mar 2025) Extracted AMESim simulation data showing hydraulic pump degradation patterns. Built data pipeline storing results on Amazon S3 for versioned access. Developed LSTM neural networks achieving 96.5% accuracy in predicting pump Remaining Useful Life (RUL). Implemented MLflow for hyperparameter optimization tracking, model performance comparison, and version control of trained models.

Working Student (July 2023 – Aug 2024) - Fraunhofer IOSB, Karlsruhe Enhanced human-machine interaction through real-time object detection systems using CNN and YOLO architectures. Contributed to predictive analytics module applying ML techniques for data-driven decisions. Collaborated in Agile environment to streamline workflows and reduce system latency. Designed ETL pipelines using Apache Airflow for preprocessing large-scale datasets and efficient ML model data flow.

Software Developer Internship (June 2021 – Dec 2021) - Cognizant, India Developed data-driven applications using SQL and Python, building interactive dashboards with enhanced visualization capabilities in Spotfire. Designed automated ETL systems improving data processing efficiency and accuracy. Implemented scheduled pipeline solutions using Apache Airflow for workflow optimization. Created analytical software tools identifying business trends and generating actionable insights for strategic decision making