Dhruvil Shah

dhruvil.d.shah25@gmail.com

University of Utah

About Me

Computer Science Enthusiast

I am a Masters student in the Kalhert School of Computing at the University of Utah. I am broadly interested in working on the intersection of data analysis and human centric computing. I am currently working on understanding how campaigns of people, and state-sponsored information operations in partcular, coordinate the creation and diffusion of strategic narratives on social media.

I received my undergraduate degree in Computer Science from SRM Institute of Science and Technology in Tamil Nadu, India

I am currently on the job market!

Experience

Research Assistant @ University of Utah
August 2022 - May 2024

Analyzed over 200M Twitter posts using Python by parsing tweets to extract and categorize URLs, identifying patterns in state-sponsored information operations.
Conducted comparative analysis of multi-variate data from human & bot users over 25 operations, providing insights into behavior patterns, and performed root-cause analysis to enhance the detection of malicious users on Twitter.
Engineered a scalable pipeline using asynchronous requests and HTTP calls to process over 50M URLs, reducing manual processing time by 40% and facilitating comprehensive.
Enhanced processing efficiency by 20% using automated scalable data solutions with shell scripting.

Data Analytics Intern @ Credohire
December 2021 - January 2022

Optimized scoring algorithms by developing a new category weighing system, dynamically adjusted for client requirements, resulting in a 7% increase in score accuracy and a 25% improvement in hiring efficiency.
Designed interactive dashboards using Python, SQL Server, and Tableau to visually present user rankings for hiring managers. Reduced initial screening time by 7 minutes per candidate.
Forecasted candidate performance trends using time-series to improve recruitment strategies

ML & Data Analytic Intern (R&D) @ Pikkal Co.
January 2021 - March 2021

Developed a speech-to-text and genre classification tool using Python and AWS Comprehend, saving over 4 minutes per podcast upload through automated processing.
Created a podcast lookup tool with efficient string parsing and ID assignments, reducing search time by 20%, and implemented a scalable pipeline using AWS Lambda.
Managed code versions using Git and followed Agile practices to support efficient development processes.

Web Development Intern @ Variant Labs
September 2020 - October 2020

Designed and developed a responsive interview portal, attracting over 1,000 monthly users, and integrated functionalities using a CI/CD pipeline.
Streamlined the recruitment process by implementing Django for efficient communication between internal databases, improving data handling efficiency.
Integrated Video Conferencing and Coding IDE using 3rd party APIs. Enhanced user interaction and experience.

Projects

History of Soccer Dashboard	Developed an interactive Tableau dashboard visualizing the history of the FIFA World Cup. Utilized Python and SQL to clean and manipulate the data before creating the dashboard, providing insights into trends and team performances. [Link to the Dashboard]
LSM-tree based Graph Store	Built a LSM-tree based, write-optimized graph store for back-end systems integrated with Compressed Sparse Row format using C++. Improved update time by 10ms. [Code][Project Report]
Cricket League Dashboard	Employed JavaScript, specifically d3.js and Python to scrape and clean data to develop an interactive dashboard visualizing IPL 2022 data. Created dynamic visualization including heatmaps, pie charts, and bar graphs. [Code][Deployment]
Visualizing Earths Mantle Convection	Gained valuable insights into the various anomalies that occur beneath the earth's surface. Assessed scalar field visualization enabled us to explore and implement a range of filters in ParaView, which proved immensely useful in visualizing the dataset. Stood 2^nd among 40 teams in a Scientific Visualization contest held at the University of Utah. [Problem Statement] [Project Files]
Fostering Tolerance through Online Collaboration	Utilized citizen science and design techniques, including UX optimization, A/B testing and prototyping with Python to investigate if online collaboration with users can foster tolerance among individuals.
Violence Detection	Trained a model to detect violence in a video using MobileNetV2 and YOLO. The model had an accuracy of 91%. [Code]
COVID-19 Tracker	Developed a website to track COVID-19 stats and act as an interface between shops selling essential goods and consumers along with a doctor-patient portal. Won the IBM special track prize at HackOn 2020. [Code]

Publications

Hydropower Plant Energy Generation Prediction Model
ICCCI 2022

Developed a project utilizing LSTMs and ARIMA to forecast energy production from Tedzani 1&2 power plants.Conducted data preprocessing and statistical analysis to ensure model accuracy. Achieved a 9% mean standard error and successfully presented findings at the International Conference on Computer Communication and Informatics 2022.
[PDF]