Krishna Sruthi Velaga | Data Scientist

Hi, I'm Krishna Sruthi Velaga

A

Detail-oriented and curious, exploring new ideas and technologies.

About

I am currently pursuing a Master’s degree in Computer Science with a focus on Data Science at Towson University, maintaining a 3.97 CGPA. My academic focus is on applying advanced techniques in machine learning, data mining, and software engineering to solve real-world problems. I am skilled in programming languages such as Python, Java, and SQL, and have hands-on experience with frameworks like TensorFlow, PyTorch, and ReactJS.

As a Graduate Assistant, I support the academic community by mentoring undergraduate students, and assisting faculty with research. Additionally, as a Lab Administrator in the Department of Computer and Information Science, I provide technical support, troubleshoot software and hardware issues, and ensure seamless lab operations while collaborating with faculty and students to resolve technical challenges efficiently.

I am currently exploring full-time job opportunities and internships where I can leverage my expertise in data science, machine learning, and software development to tackle complex challenges and contribute to impactful projects.

Experience

Department of Computer and Information Sciences, Towson University

Graduate Assistant

Created and maintained a detailed spreadsheet of 260 U.S. universities to support faculty research.
Conducted lab sessions, guiding over 60 students per semester in understanding advanced programming concepts.
Debugged coding issues for students, ensuring completion of lab and strengthening their problem-solving skills.
Tools: C++, Debugging

Aug 2024 - Present | Towson, MD

Department of Computer and Information Sciences, Towson University

Lab Administrator

Provided comprehensive software troubleshooting, including the installation and configuration of programming IDEs and virtual machines like VMware and VirtualBox.
Resolved numerous hardware-related issues, ensuring all computer stations and peripherals were fully operational for students and faculty. environment
Offered clear, step-by-step instructions to students on using technologies, enhancing their understanding of lab technologies and reducing the recurrence of technical issues.
Tools: Troubleshooting, Technical Support, Software Installation, Virtual Machines (VMware, VirtualBox), Programming IDEs (PyCharm, Visual Studio Code)

Aug 2023 - Present | Towson, MD

Projects

                
Optimizing Large Language Models Assisted Smart Home Assistant Systems at the Edge
                  A pilot study to evaluate the real-time performance of the edge deployment of LLMs on practical resource-constrained devices.
                
                Accomplishments
                Tools: Raspberry Pi 5, Home Assistant, Python, Pandas/NumPy
                  Integrated lightweight LLMs (Home-1B, TinyHome, TinyHome-Qwen, StableHomeLM) with Home Assistant on Raspberry Pi
Achieved an 82% latency reduction (45.1s to 7.9s)
Create a synthetic dataset replicating home device interactions.
Watch on YouTube
                
Brain Tumor Detection
                  A CNN model for brain tumor classification, analyzing the impact of dataset poisoning on performance
                
AccomplishmentsTools: Python, TensorFlow, CNN, NumPy, Matplotlib
                  Built a CNN model to classify brain tumors from MRI images.
Executed a label flipping attack to simulate dataset poisoning and assess model vulnerability.
Analyzed performance using metrics like accuracy and loss to understand the impact of clean vs. poisoned datasets.

Analysis of 911 Calls for Detroit and New York City
                  Analyzed 911 call patterns to optimize emergency response times and resource allocation using predictive modeling.
                
AccomplishmentsTools:  Python, SARIMAX, XGBoost, DBSCAN, K-Means, Pandas, NumPy, Matplotlib
Analyzed 911 call data to identify emergency hotspots and improve public safety.
Implemented predictive models (SARIMAX, XGBoost, DBSCAN, K-Means) to forecast emergency incidents.
Provided insights to optimize emergency response times and improve resource distribution.

Inventory Management System for Local Restaurant
                  Developed a desktop application to streamline restaurant operations, including inventory and employee data management.
                
                Accomplishments
                Tools:  Java, Swing, SQL, Triton, Git
                  Created a desktop application to manage restaurant inventory and employee records.
Implemented secure admin and employee modules with authentication features.
Hosted the database on Triton and ensured data security and reliability.
Collaborated with a team of five to design a user-friendly interface and database structure.

Malicious URL Detection using Machine Learning
                  A Chrome extension to detect and alert users of malicious websites, enhancing online security.
                
AccomplishmentsTools: Python, BeautifulSoup4, GoogleSearch-Python, Scikit-learn, Chrome API
                  Created a Chrome extension to notify users when accessing potentially malicious websites.
Enhanced model accuracy by integrating content-based (e.g., HTML structure, embedded scripts) in-addition to lexical, and host-based features.
Leveraged Python libraries such as BeautifulSoup4 and Scikit-learn to build and train the detection model.
Presented the project at ICRASTEM 2023

Event Management 
                  A web application to streamline the event planning process with services like venue booking, decor, and logistics management.
                
AccomplishmentsTools:React.js, Node.js, EmailJS, HTML5, CSS3, Material-UI
                  Allows users to submit event planning requests through a contact form.
Provides a user-friendly interface to explore various event services.
Handles corporate events, social gatherings, weddings, and more.
Integrates EmailJS for direct client-planner communication without a server.
Uses React.js for the frontend and Node.js for backend API handling.

Blood Bank Management System
                  A web-based application that connects patients, donors, and blood banks to streamline blood requests and donations.
                
AccomplishmentsTools: Django (Python), HTML, CSS, MySQL
                  Patient Module: Allows patients to register, log in, request blood, and view request history.
Donor Module: Enables donors to register, donate blood, request blood, and view donation history.
Admin Module: Admins can manage blood stock, approve/reject donations and requests, and monitor available units by blood type.
Centralized platform that improves the efficiency of blood request and donation processes.