Hi, I'm Krithi! Welcome to my corner of UTCS. I am deeply passionate about computer science, and especially fascinated by the synergy between machine learning and systems engineering. The potential to create systems that not only operate efficiently but also learn and adapt has always inspired me. I find joy in exploring how OS-level optimization can support scalable machine learning models, enhancing both performance and flexibility. This intersection excites me because it allows for innovative applications, from efficient resource management in real-time systems to the development of intelligent, autonomous processes. Pursuing knowledge in these fields drives me to push the boundaries of what's possible in computing.

My goal, not just in computer science, but with life in general, is to make the world a better place by creating tools and solutions that have a lasting positive impact. I believe that technology has the power to break down barriers, improve access to resources, and solve problems that might otherwise feel insurmountable. Through my work at UT and beyond, I hope to contribute to projects that make people’s lives easier and open up new opportunities for connection, learning, and growth. Whether it’s through designing more efficient systems, developing intelligent applications, or finding new ways to use data to inform decision-making, I’m driven by a commitment to use my skills for good and to leave a positive mark on the world around me. If you're interested, please check out my projects below!

I'm currently working on...

Research under UT Austin's Astronomy department with Dr. Shyamal Mitra! – I am actively involved in astronomy research as part of the College of Natural Sciences Research Initiative, where I develop and test machine learning models to classify low-surface brightness galaxies. My work involves using advanced techniques like XGBoost and neural networks, alongside tools like TensorFlow and Astropy, to streamline galaxy type identification.

A machine learning fellowship with Google via my scholarship program, Break Through Tech! – I am working on optimizing predictive models for YouTube video performance. This involves using early engagement metrics, video metadata, and trends to improve predictive accuracy. I am also implementing efficient data preprocessing pipelines to integrate external data sources seamlessly into classifiers and regressors.

Research under the LDOS team with Dr. Aditya Akella! – I am exploring the application of graph neural networks to predict operating system resource management needs based on changing environments. At LDOS, my goal is to enhance the adaptability and efficiency of OS resource management by creating models capable of handling dynamic, real-world scenarios.

My past work...

Handwriting Based Parkinson's Disease Diagnosis – During this research project, I developed a neural network to determine the likelihood of a patient having a positive diagnosis of Parkinson's Disease by analyzing a handwriting sample in conjunction with the timestamps of each pixel while writing. This is one of my favorite projects that I've worked on, as it allowed me to apply machine learning to a real-world medical problem with the potential to improve early diagnosis and patient outcomes.

Determining Soil Organic Matter – My first foray into machine learning, years ago! This was a research project where I developed a regression model in combination with a mobile app to predict the amount of organic matter in soil, with the hopes of reliably testing soil samples for their compositions just by using a phone camera.

Other cool projects which I hope to come back and refine when I have time sometime in the future...

Treatsi – Did you know that small businesses make up 99% of U.S. businesses? They provide over half of all jobs, yet they struggle with funding, visibility, and survival, with nearly 45% failing within five years. Treatsi is a web application that addresses this by spotlighting local cafes and restaurants, offering customers a unique marketplace and incentivizing repeat visits through a rewards system, while empowering minority-owned businesses and providing entrepreneurs with valuable analytics to strengthen their models.

ECLAIR Raspi Project! – A few semesters ago, I led the ECLAIR Robotics Beginner's Track, and I'm super proud of what the team was able to create despite having little to no experience with full-stack development. The end result was a Raspberry Pi that mimicked a home automation system: with a mobile app on our phone, we could say "Turn on the light," and a lightbulb on the Raspi would light up accordingly. It was such a fun project and I loved teaching and helping my team.

Determining Abnormal Breathing Patterns for COVID-19 Diagnosis – The aim of the study was to develop a non-invasive app that allowed users to assess their respiratory health by analyzing audio waveforms of their breathing patterns, providing a simple tool for early COVID-19 diagnosis. I compared the app's results with FDA-approved pulse oximeters, and the research demonstrated the potential to offer a reliable self-assessment method, expanding knowledge in computerized respiratory sound analysis for better patient accessibility.