Raj Rajeshwari Prasad

Responsible for building a complete Machine Learning/ Deep Learning pipeline for the purpose of Predictive Maintenance in order to prevent downtime of machines.

Explored the concept of Self-Supervised Representation Learning to enhance the quality of the learned embedding space and reduce the dependency on target labels. The learned embeddings were further used to build a classification model for classifying surface defects of Body-in-Whites into respective defect classes. Further, analysed the nature of the learned representation space and plotted saliency maps to introspect the behaviour of the DNN.

Developed a pipeline for the classification of surface defects for digitalizing Quality assessment of Body-in-Whites. The core tasks involved data analysis, instance segmentation, data pre-processing, developing Machine Learning/Deep Learning models and building a test pipeline for defect assessment in the production facility.

Accomplished several projects under the department of Research & Development to develop Wire Wound/Grid Resistors and Industrial Heaters for clients in the Power and Automotive sectors. The core tasks were designing new products, performing stress analysis (CAE), prototype development and testing. Additional role incorporated auditing the company’s manufacturing unit to ensure product quality.

Specialization in concepts of Artificial Intelligence like Machine Learning and Deep Learning Master Thesis: Classification of Surface Defects of Body-in-Whites using Contrastive Learning. Academic Projects: 1. Reinforcement Learning for Scheduling Problems. 2. Sequence Pattern Mining. 3. Entity Resolution for Heterogeneous Data. 4. Book Genre Identification. 5. Android Application for Monitoring Monthly Expenses. 6. Suggesting Measures to Enhance the Safety of a Traffic Node with Simulation.