Experience

Research Intern

SERL Lab | DAIICT Gandhinagar
May 2024 - July 2024 (2 months)

Applied zero-shot and few-shot learning prompts through the Gemini API to generate test cases for regression testing, improving efficiency and coverage on small-scale software projects.
Integrated CFG analysis to identify critical execution paths and validate test cases on smaller code examples, optimizing test coverage and ensuring code reliability.
Used Regression prompt to classify test cases as outdated, relevant, or new, improving test suite accuracy and reducing manual effort for smaller-scale test scenarios.

Bolbala Trust | Rajkot, Gujarat
December 2022 - January 2023 (2 months)

Direct Aid and Community Support: Provided essential resources, including food, shelter, and clothing, to underprivileged individuals residing in slum areas, directly addressing their immediate needs.
Operational and Administrative Support: Contributed to the trust’s operational efficiency by assisting with data entry, distributing promotional materials, managing donations, and supporting membership activities.

My Projects

Regression Test Selection Using Pre-Trained Large Language Models

Abstract:
This paper explores the possibility of automated generation of regression test sets using pre-trained Large Language Models (LLMs). We apply augmenting prompts with the test cases derived for the Code Under Test (CUT) and the identification of the regression test set. We leverage LLMs to create exhaustive test cases and use Control Flow Graphs (CFGs) to assess the quality of generated test cases from the LLMs. A dual classification process allows us to identify which of the test cases are outdated, relevant, and new, providing a means for both efficient and effective regression testing.
Our preliminary analysis results show that the regression test set generated by the pre-trained LLMs, on average, have 96.4% precision and 85.1% recall, indicating LLMs’ applicability in regression test selection.

This research project was done under the guidance of Professor Saurabh Tiwari. The paper is submitted in ISEC 2025 Conference.
Skills: Python, Artificial Intelligence, Software Testing, Google Gemini

Used Vehicle Sales: Trends and Patterns Through Data

Performed Exploratory Data Analysis, preprocessing, cleaning, and data visualization on the Vehicle Sales dataset.
Collected a dataset of 700 cars from Cars24.com using web scraping with Beautiful Soup, then cleaned and organized the data to fill in missing details and ensure accuracy.
Used libraries like Pandas, Seaborn, and Matplotlib to explore the data, creating graphs and charts (scatter plots, histograms, etc.) to identify patterns and improve the selection of important features.
Developed models to predict car prices and fuel efficiency using techniques like Random Forest, Linear Regression, and XGBoost.
Improved the accuracy of these models, achieving a high R² score of 0.976 by fine-tuning the important features.
Skills: Pandas, Scikit-Learn, Linear Regression, XGBoost, Statsmodels, Seaborn, Beautiful Soup

Crisis Data Summarization for Emergency Management

Developed a system to create concise summaries of crisis-related data for use by emergency responders. Worked with a multi-stream dataset that included social media posts, news articles, and other updates.
Used advanced tools like BM25 and BERT to filter and rank key facts based on relevance. Implemented both extractive and abstractive summarization techniques to balance accuracy and readability.
Skills: Information Retrieval, Python, Deep Learning, Machine Learning, nltk

Learning and Exploring 🚀

I’m constantly expanding my skillset and exploring new technologies. Currently diving into:

Cloud Computing: AWS ☁️ and Azure 🌁
Artificial Intelligence & Machine Learning: AI 🤖 and ML 🧠
Web Development: Java ☕, JavaScript 📜, React ⚛️, Node JS, Express JS, Angular JS
API Development & Testing: Postman 📬
Open Source Contributions: Eager to contribute to open-source projects! Looking for opportunities to collaborate and learn. Feel free to connect! 🤝

Let’s connect! ➡️ LinkedIn

Github