Data Entry Encoder

Shruti Shah

919-***-****• *******@****.*** • linkedin.com/in/shrutishah-1998 • Raleigh, NC 27606

EDUCATION

North Carolina State University Jan 2021 - Dec 2022

Masters in Electrical and Computer Engineering 3.84/4.00

Coursework: Neural Networks, Computer Vision, Digital Imaging Systems, System Control Engineering, Digital Communication

MIT College Of Engineering, Maharashtra, India Jun 2016 - Jun 2020

Bachelors in Electronics and Telecommunication Engineering 9.23/10.00

Coursework: Color and Image Processing, Audio and Video Processing, Digital Signal Processing, Data Structures and Algorithms

SKILLS AND COMPETENCIES

Programming languages: Python, PySpark, MATLAB, C++, JAVA, NodeJs, Angular, SQL

Web Development Framework: HTML, CSS, PHP

Operating Systems: Windows, Linux

Simulation/ Visualization Tools: Simulink, PowerBi, Tableau, QlikView

Machine Learning tools and Framework: OpenCV, NumPy, Matplotlib, Scipy, Keras, TensorFlow, Pandas

WORK EXPERIENCE

Murano Corporation – Data Science Intern(Raleigh) Jan 2022 – Present

Created framework for data acquisition, dissemination, and prediction cyber security tablet with advanced capabilities on the edge by integrating with Electronic Control Modules via CANBus and SCADA (and Modbus, Profibus etc.)

Designed software for lifecycle asset management including managing designs and models, computer vision, Natural Language Processing (NLP) to ensure spare and replacement parts availability, provide relevant repair and maintenance instructions and optimized maintenance schedules

Modeling and simulating nuclear power plants to detect and diagnose fault conditions, predicting future failures and remaining useful life using Logistic Regression, LSTM, SVM, CNN and Exponential degradation expecting a reduction of annual inspection and maintenance labor costs by 50-75%

Suyog data Consultancy-Software Trainee(India) May 2019 – Dec 2019

Developed a database management system for jewelry shops which reduced time loss in data entry by 80%

Provided better analytics by integrating vendor, inventory, sales and financial management systems

ACADEMIC PROJECTS

Terrain Classification from time-series data Python, Keras Aug 2021 – Dec 2021

Preprocessed data using down-sampling techniques and created custom time window frames

Classified different terrains by IMU sensor(acceleration and gyroscope) data by testing it on models like Conv1d, CNN, RNN+ LSTM and dense layers in keras and achieved an accuracy of 84% for best RNN+LSTM model

Image Caption Generator using Deep Learning Python, Keras, TensorFlow Aug 2021 – Dec 2021

Trained the Imagenet dataset on CNN model called Xception (encoder) for image feature extraction. Image Caption was generated by feeding these extracted features to LSTM and GRU(decoder)

Compared various encoder decoder models to analyze how each component influences caption generation and observed LSTM model performed better than GRU

Facial Image Classification using Statistical Model for Computer Vision Python, OpenCV, Keras Jan 2021 – April 2021

Generated train and test Dataset from FDDB dataset with Intersection over Union (IoU criteria)

Trained Single Gaussian, Mixture of Gaussian, T-distrubtion and Factor analyzer on 1000 face and 1000 non-face images

Evaluated and compared the models based on ROC curve and misclassification rate

Implemented a CNN based classifier to classify images. Tuned the hyper parameters by babysitting the training of the network and employing a course and fine search for the parameters on log scale

After hyper tuning of the parameters accuracy of 96.48% and testing loss of 0.077 was obtained

Maintenance of Automobiles by Predicting System Fault Severity Machine Learning Jan 2020 – May 2020

End to end open source predictive maintenance solution to predict severity of faults in a car with help of historical and real time sensor data using IoT and machine learning.

Sensor data is collected from a Suzuki Swift Model and classifiers like Logistic Regression, Random forest and Gradient Boosting Tress are used to train the data with imputed faults.

An end to end OBD data to user dashboard pipeline is proposed with final predicted faults visible on a real time dashboard.

Gradient Boost gives the best performance result with f1 – score as 0.947, AUC as 0.996 and accuracy as 94%

CONFERENCE PUBLICATIONS

(Research publication)”Maintenance of Automobiles by Predicting System Fault Severity using Machine Learning” -

2nd International Conference on Sustainable Communication Networks and Applications & Springer (ICSCN 2020)

Contact this candidate