Sona K
- Research Program Mentor
MEng at Cornell University
Expertise
Software Engineering, Machine Learning, Data Science, Applied AI/RAG Systems, Full-Stack Development
Bio
I’m a computer science graduate student at Cornell University focused on machine learning, data science, and applied AI systems. I have experience building AI and software projects across areas like full-stack applications, machine learning models, RAG systems, and product focused technical tools. My academic and professional interests sit at the intersection of software engineering, AI, and product development, especially building technology that solves real-world problems. Outside of academics, I enjoy traveling to new places, trying different restaurants, and drawing. I also like building side projects, staying active, and spending time with friends. I’m excited to mentor students who are curious about computer science, AI, and turning technical ideas into real products.Project ideas
Building a Campus Club Event Finder
In this project, you will explore software engineering and product design by creating a web app that helps students find events from different campus clubs in one place. You could gather sample event data from public club websites, Instagram posts, or a manually created spreadsheet with fields like event name, date, location, category, and RSVP link. You will learn how to define a real user problem, design a simple interface, organize data, and build features such as search, filters, and saved events. The final outcome could be a working prototype, or technical write up.
Evaluating AI Answers for SAT or AP Questions
In this project, you will explore applied AI by testing how well AI tools answer a set of SAT style reading questions or AP practice questions. You will gather 30–50 public practice questions, prompt an AI system to answer them, and compare the responses against answer keys and explanations. Students will learn about prompt design, accuracy scoring, hallucinations, and how to evaluate AI systems beyond whether an answer “sounds right.” The final outcome could be an evaluation report, comparison analysis, or testing framework.
Predicting Diabetes Risk from Health Indicators
In this project, you will explore machine learning and data analysis by building a model to predict diabetes risk using a public dataset such as the CDC Diabetes Health Indicators dataset. You can analyze features like age, BMI, physical activity, blood pressure, cholesterol, smoking history, and general health ratings. You will learn how to clean data, create visualizations, train a predictive model, evaluate performance, and discuss limitations around fairness, privacy, and real world medical use. The final outcome could be a Jupyter notebook, research paper, or presentation with charts and model results.