Working on two problems, one of which is funded by Epitel Inc, a leading enterprise in seizure monitoring and prediction. I have been involved in the following projects:• Development of a full framework to detect and predict different types of epileptic seizures using single channel EEG signals and machine learning algorithms. The EEG signals are obtained from a novel state of the art wearable EEG sensor called EpilogTM• Development of a fast-online unsupervised learning framework to detect and characterize damages and defects in Aluminum plates using machine learning algorithms• Mentored one undergraduate student researcher and a high school student in the field of Structural Health Monitoring and EEG signal analysis respectively

Teaching Assistant for EECS 103: Special ProjectsGrader for ECE 499: Statistical Signal ProcessingTeaching Assistant for ECE 351: Signals and Systems ITeaching Assistant for ECE 342: Junior DesignTeaching Assistant for ECE 451: Digital Signal ProcessingTeaching Assistant for ECE 202: Electrical Fundamentals II

Worked on using raw accelerometer data obtained from a wearable watch and machine learning algorithms to objectively estimate fatigue in patients with Lupus and Sjogren’s syndrome compared to demographically matched healthy volunteers.• Extracted several activity, sleep, and circadian rhythm-based features from accelerometer data obtained using a wearable watch• Implemented a feature selection algorithm (based on mutual information) and SHAP.• Explored several different classifiers such as XGBoost, ExtraTrees Classifier, Random Forests, etc• Explored the importance and the reasoning behind defining a fatigue threshold.• Presented a poster and was the first author of a conference publication. (IEEE EMBC 2024)

Worked on a multi-university project funded by the US Department of Transportation. The following research was undertaken:• Involved developing multi-modal NDE techniques for the detection and characterization of defects in long complex pipelines• Performed numerical modelling and simulated corrosion pits and cracks in a pipeline and study the guided wave propagation in these pipelines• Developed different machine learning based classification algorithms for classifying the simulated defects• Project Link: https://primis.phmsa.dot.gov/matrix/PrjHome.rdm?prj=840Developed a framework to characterize the material properties of composites using Deep Learning algorithms for my Master’s thesis. The following research was undertaken as part of this research:• Developed a COMSOL based 2-D models as frequency domain based Spectral FEM composite models• Set up guided wave simulations in the developed models to help characterize the material properties• Developed Deep Supervised learning algorithms like Convolutional Neural Networks, Long-Short Term Memory networks and Dense MLPs to perform regression and identify the material properties of different composites

Teaching Assistant for ECE303: Electronics Laboratory

Teaching Assistant for ECE303: Electronics LaboratoryTeaching Assistant for ECE415: Computer Aided Manufacturing

Worked on a project funded by Magna International Inc to detect defects in bonded thermoplastics lift gates and carbon fiber composites using NDE techniques. The following research was undertaken:• Developed a methodology to characterize adhesion of adhesively bonded thermoplastics heavily used in the automobile industry• Performed several ultrasonic testing experiments to determines the effect of surface treatment on the adhesion quality. A-scans, Immersion scans and air coupled scans were performed as part of this.• Experimentally obtained data was processed using signal processing algorithms to determine and characterize adhesion qualityWorked on the ultrasonic material characterization of Poly-Dimethyl Siloxane (PDMS), a soft polymer. The effect of adding silicone nanoparticles to the PDMS matrix was studied. Conducted various experiments to experimentally determine the material properties of PDMS and PDMS based polymer composites.I also worked on NLP models, to determine the sentiment of various tweets from the GOP debate twitter dataset. I utilized smart versions of LSTMs that greatly reduces the total number of learnable parameters and computation time

Teaching Assistant for ECE303: Electronics LaboratoryTeaching Assistant for ECE415: Computer Aided Manufacturing

Worked on my undergraduate thesis that involved the development of a convolutional neural network-based algorithm to land an Unmanned Aerial Vehicle on a moving Unmanned Ground Vehicle.

• Assisted in the building of an obstacle avoiding UGV• Assisted in the development of a robotic arm that can grasp spanners and go to a location to do a related job.• Understood the basics of Deep Learning and Robot Operating System (ROS)

I worked on interfacing basic sensors and different communication modules with Arduino. Also learnt how to 3D print basic structures. Part of the job was also to write about the same in the company blog. The links for a few of them are attached below: • http://invent.module143.com/analysing-mq-5-gas-sensor-using-thingspeak/ • http://invent.module143.com/arduino-basics-playing-around-with-adxl335/ http://invent.module143.com/battery-tester-using-leds/

.