Projects:• Learning for Edge Devices: Working on deep learning model compression and optimization algorithms for deployment of time series classification models on edge devices with focus on tiny devices and sensor nodes.• Hybrid Learning: Working on approaches that make combined use of traditional signal processing techniques and deep learning methods for time series / sensor data classification tasks.

Projects:• Learning for Edge Devices• Hybrid Learning• Sensor Informatics Workbench: Worked on design and development of an automated workbench for generating optimized machine learning models for time series classification. This includes preprocessing raw data, feature generation, identification or selection of useful features, training and tuning of classification models. 1 granted patent.

Projects:• Sensor Informatics Workbench

The project was under PT&D Department, Electrical Division of MECON Ltd. The work involved:• Study and preparation of Techno-Economic Feasibility Report (TEFR) of a steel plant as regards to its electrical systems.• Transformer sizing calculations, that is, determining the proper capacities and ratings to use in a given industrial application.