With experience of a Data Engineer, I ventured out in the domain of applying deep learning and machine learning to my research work at the University of Washington. This was my main motivation which led me to my passion in Data Science.
June 2018 - Current
• Built a tool to detect early crashes using Machine learning in Concur Mobile App and deployed it in AWS reducing detection time by 48%.
• Led the initiative for development of two insights tools (dashboards) using Python/HTML which automatically generates top topics from daily customer/user feedback used by all the product managers and TPMs in the company which helps in directly identifying key issues in products and user experience.
•Identified the key metrics and built dashboards on Tableau which helps in continuous monitoring of P2s/P1s.
November 2016 - August 2017
• Developed two full modules using SQL, ETL where Big Data (more than 2 million records of raw data) from multiple sources is cleaned, formatted and transformed and then, aggregated in the data warehouse within a period of 2 months. This ensured the timely product release for the client.
• Analysed and interpreted data reports of the aggregated data and visualised data on dashboards using Tableau, IBM Cognos which helped for driving data driven business solutions and strategic advancement of finances of the client by 11.75%.
• Gave regular presentations about the module to the client and team for streamlining the processes.
June 2018 - Current
• Identifying the features which affect the rate of heart attacks in the Framingham Heart Study using Machine learning (PCA)
• Theme identification of speech transcript data of Alzheimer's disease using machine learning by implementing feature reduction algorithms such as NMF(Non-Negative Matrix Factorization) and LDA (Latent Dirichlet Allocation) under Prof. Reza Hosseini.
• Performing sentiment analysis on the Parkinson's disease data using deep learning (LSTMs).
February 2018 - June 2018
• Sentiment analysis of Ebola disease and feature reduction using machine learning
• This is a research work under Prof. Benjamin Althouse (Institute for Disease Modeling) which concentrates at text mining and performing sentiment analysis using Python to understand the social drivers of the Ebola epidemic.
• The final step is to improve the accuracy and ROC, eliminating redundant features using feature reducing algorithms (PCA).
September 2017 - June 2019
• Completed coursework: Deep Learning for Computational Scientists, Artificial Intelligence for Engineers, Data Science I (Predictive modeling and causality), Data Science II (Machine Learning Algorithms),Data Science III (Scaling and deep learning), Business Intelligence Systems(Data Analysis and Visualization)
August 2012 - June 2016
• Completed coursework: Random Signal Analysis(Statistics), Discrete Time Signal Processing (Time Series Analysis), Object Oriented Programming, Structured Programming Approach, Data Compression and Encryption
Husky 100 Award - 2019
Honored for being selected among the best 100 students for 2019 in University of Washington
What is Husky 100? The Husky 100 recognizes 100 UW undergraduate and graduate students who are making the most of their time at the UW.
Published school website: https://www.washington.edu/husky100/#name=prithvi-shetty
Best Intern Award - 2018
Completed 5 intern projects within a timeframe of 10 weeks. Implementation of deep learning and machine learning for text classification and topic detection in text.
Published blog: https://blogs.sap.com/2018/08/22/winners-circle-sap-interns-outshine-with-talent/
First place
Our product "Retail'ored" was targeted to reduce the number of returns in digital commerce with a three-step approach. The solution combined user experience research and data science in a constructive way.
The three step approach consists of a ranking and recommender system using machine learning, a deep learning based sentiment/theme classification algorithm and a mixed reality experience that enabled the users to experience products before buying.
Github link : https://github.com/shettyprithvi/Hacksgiving
First place
Won the first place at Seattle. Also, nominated to represent in the global round amongst the top 100 teams in the world.
Name of product prototype : "Future Suture". The idea is to build a machine learning model for predicting climate changes of regions across the globe and visualize it using Virtual reality.
The end goal was to create awareness about the effects of human activities on our planet.
Project link: https://2018.spaceappschallenge.org/challenges/what-world-needs-now/globe-observer/teams/stardust-explorers/project
First place
Idea : Reduction of waste and eradication of starvation by leveraging data.
Product : Mobile application which routes data and acts as a medium between food providers such as restaurants who have excess food and the needy people who are seeking food. Also, collaborates all the data of other available free food sources such as food banks/pantries in the application.
Name of product prototype : "FoodCast"
Machine learning/Deep learning
Given AirBnb data, I apply deep learning using Recurrent Neural Networks (LSTMs) for classification of text reviews to visualize impact on price. Also, the highly desired amenities are found using scattertext created in Python. For further details, please visit the Github link.
Github link : https://github.com/shettyprithvi/Text_classification_deep_learning
Deep learning
Implementing Convolutional Neural Networks to identify metastatic cancer in small image patches taken from larger digital pathology scans.
Github link : https://github.com/shettyprithvi/Cancer-detection-image-classification-Convnet
Deep learning
This project aimed at Use object detection on a car detection dataset and deal with bounding boxes using Convolutional Neural Networks implementing YOLO algorithm. For further details, please visit the Github link.
Github link : https://github.com/shettyprithvi/Convolutional_Neural_Networks
Artificial Intelligence
The shortest path is found by the AI Robot by finding the best and the shortest path using A-star algorithm. The heuristic is the Euclidean distance between the robot and destination. For further details, please visit the Github link.
Github link : https://github.com/shettyprithvi/Artifical_Intelligence_Pathfinding_Robot
Artifical Intelligence
The game is played by AI designed by me against the human player. The implementation is done using Minmax algorithm with Alpha-Beta pruning. For further details, please visit the Github link.
Github link: https://github.com/shettyprithvi/Artificial_Intelligence_Mancala_game
Data Engineering/Business Intelligence
This project involved end-to-end designing and building a data warehouse from raw data using SSIS, SQL Server. The final step was to visualize data on dashboards of Tableau and provide recommendations based on the visualizations. For the tableau visualization, please visit the Github link.
Github link : https://github.com/shettyprithvi/Data-visualization-of-sales-data
deeplearning.ai
The deep learning specialization helped me a lot to learn and apply Deep Neural Networks, Convolutional Neural Networks and Recurrent Neural Networks.
Certificate link: https://www.coursera.org/account/accomplishments/specialization/24N8D29MG8LK
Stanford University
This course by Andrew Ng helped me in understanding the regression analysis and learning algorithms (Recommender systems, Neural networks, Clustering by K-means, Support vector machine, Logistic regression) for supervised and unsupervised learning models.
Certificate link: https://www.coursera.org/account/accomplishments/verify/358Q2VAWKLSG
University of Michigan
This course helped me understand the core data structures of the Python programming language. Also, I explored how to use the Python built-in data structures such as lists, dictionaries, and tuples to perform increasingly complex data analysis
Certificate link : https://www.coursera.org/account/accomplishments/verify/WNNXEMRYGHWQ
Project steps: PCA, Image Compression, Optical Character Recognition, RDBMS
The project is a mobile application which implements image processing using PCA (Principal Component Analysis) and OCR (Optical Character Recognition). SQL is used for getting details of scanned number plates from the database.
Published technical paper link: http://www.ijesi.org/papers/Vol(5)10/B05107011.pdf
I would love to speak with you. Make sure you drop your email and your name along with a brief message. Thank you!
prithvi1 [at] uw [dot] edu
Seattle, WA
US