Hi I'm Saisha Shetty, a second-year Master's student pursuing an MS in Computer Science at University of California, Davis. I currently work as a Graduate Student Researcher, where I focus on annotating radiological clinical notes using large language models. This research aligns with my keen interest in the intersection of AI in health and ML systems, as I aim to contribute to advancements in healthcare through technology.
With a passion for Deep Learning and Machine Learning, my journey into this field began in 9th grade when I first encountered Sofia, the robot granted citizenship. Since then, I've delved into Artifical Intelligence courses and projects, finding inspiration in the transformative power of technology. Proficient in Python, C++, and C, I believe in fostering rapid and sustainable growth through innovation, and I enjoy sharing insights through blogs on Machine Learning topics.
Completing my Masters in University of California, Davis.
Secured my Bachelor of Technology degree in SRM Institute of Science and Technology, India.
Worked as a Health data analyst and conducted through Exploratory data analysis on annotated clinical reports, create an intelligent data import module, optimizing SDoH dataset detection and ensuring seamless integration with a focus on simple systems for improved workflow.
Worked as a research intern at Energy in Cities Lab and assisted on the project "Applying big data analysis and/or machine learning approaches to building energy use" by Dr Ralph Evins. The aim was to find different factors that affect carbon emissions, heat, and boilers with data from various real-life buildings worldwide. Optimised the model by working with energy hub systems. While this program provided me with an extensive understanding of multiple data analysis techniques and technologies used in energy systems, it also made me eligible for the Globalink Graduate Fellowship of $15,000 CAD financial support at any of their partner institution for full master's or Ph.D. programs
To write code to collect, crunch and analyse data from various internal and external sources. To create linkages between various data within our business intelligence software to enable predictive modelling and trend analysis.
Developing a LLM-powered tool to annotate radiology reports, capturing relationships by fine-tuning models like LLaMA and Mixtral, aimed at enhancing clinical data accuracy and streamlining workflows.
Exploring the integration of gamification into AI education through serious digital games, which impart training objectives with game-like elements, gaining relevance in AI learning. The aim of this study is to comprehensively analyzes serious games in AI, machine learning, and data science, aiming to discern their capabilities, identify gaps, and enhance problem-solving experiences. We aim to contribute insights advocating for equitable AI education, making it accessible and engaging for learners from diverse backgrounds.
Worked on accurately identifying orthopaedic implants, specifically wrist implants, using transfer learning models like ResNet, VGG, and MobileNet to classify the makes and models of these implants from X-ray images. This is crucial for revision surgeries, minimizing surgical delays and ensuring better patient outcomes in the growing field of wrist replacement surgeries.
Working as a Teaching Assistant in the Spring Quarter of 2024 for the course 'ECS 36-B Software Development & Object-Oriented Programming in C++' under Professor Felix Wu. Responsibilities include holding office hours, conducting discussion sessions, and grading student papers and assignments.
Worked as a Teaching Assistant in the Winter Quarter of 2024 for the course 'ECS 154A- Computer Architecture' under Professor Matthew Farrens. Responsibilities include holding office hours, conducting discussion sessions, and grading student papers and assignments.
Worked as a Teaching Assistant in the Fall Quarter of 2023 for the course 'ECS 154A- Computer Architecture' under Professor Matthew Butner. Responsibilities include holding office hours, conducting discussion sessions, and grading student papers and assignments.
Began volunteering at the “AARAMBH” an NGO which works for the most marginalised families living in urban slum communities. I conducted lessons for fifth-grade students here to help develop spoken English skills and work especially with female students to help their parents secure a better education for them.
Check out my projects on Github
Toxicity Disengagement analyzed commit message toxicity from GitHub using NLTK, OpenAI Moderation API, and Hugging Face classifiers, showing positive sentiment boosts developer engagement.
SmartStudy aims to enhance student learning by comparing Retrieval-Augmented Generation (RAG) models, evaluating the accuracy and insights of ChatGPT-4, Langchain with GPT-4, and LLAMA using BLEU scores to improve student engagement.
An implementation of the research paper "Retina Blood Vessel Segmentation Using A U-Net Based Convolutional Neural Network"
Built a PaletteNet network, which recolors an image according to a given target color palette. It takes two inputs a source image to be re-colored and a target palette. Human experts with a commercial software take on average 18 minutes to recolor an image, while PaletteNet automatically recolors plausible results in less than a second.
This project provides a review system that aims to use sentiment analysis to predict the nature of a review entered by the user. It also has a movie recommender system that recommends a set of movies that the user would like to watch, based on a movie that the user enters. In order to help the user we have made a bot - ROBO to assist the user.
Contributed to analyzing the sentiments from the client using NLP Libraries, which is part of a product "FREN" meant to strengthen the relationship between a Therapist & its Client. The product channels the Client's emotions and feelings anywhere, anytime, providing a detailed analysis tailored to the Client itself.
Contributed to the project by implementing caffemodels to detect facemasks, leveraging them to calculate social distances. It helps prevent hotspots by monitoring COVID prevention guidelines using existing cameras and servers, reducing transmission risk and enhancing productivity.
Published an introductory article on edge detection algorithms like Canny and Sobel on Medium.
Published part 2 of the article, delving into the code implementation of edge detection algorithms such as Canny and Sobel.
Published a research paper on identifying the make and model of wrist implants using deep learning techniques.
Awarded $1500 as ISACA Sacramento Chapter Scholarship Recipient.
Awarded GHC23 scholarship for tech commitment and diversity dedication.
Selected for UC Davis' International Women's Day campaign.
Secured 2nd Place (out of 98 finalists) at hackCBS3.0 with a deep learning solution.
Secured 4th place at HackTheMountain with a machine learning innovation.
Elected as Operations Leader at SRM's Machine Intelligence Community.