Projects | Nitish Ramaraj

Brain Tumuor Detection.

Sat, 27 Apr 2024 00:00:00 +0000

This project aimed to detect brain tumors from MRI images using a TensorFlow-Keras implementation of the VGG19 architecture. Leveraging a dataset of 10,000 augmented MRI images, the model achieved an accuracy exceeding 80%. Through meticulous preprocessing, augmentation, and training, the VGG19-based model demonstrated robust performance in distinguishing between tumor and non-tumor regions in brain scans. With further refinement and potential deployment in clinical settings, this approach holds promise for enhancing the efficiency and accuracy of brain tumor diagnosis, showcasing the potential of deep learning in medical imaging tasks.

AI-Powered Patient Summary Generator

Wed, 27 Dec 2023 00:00:00 +0000

Developed a full-stack Next.js application that enables healthcare providers to upload medical records for AI-generated summary reports. Integrated OpenAI’s GPT API to analyze patient vitals and identify serious abnormalities, allowing doctors to quickly access essential patient information.

Sell-Easy

Tue, 27 Sep 2022 00:00:00 +0000

Sell-Easy an innovative platform developed to facilitate the buying and selling of pre-owned items within the campus community. With a focus on sustainability and affordability, Sell-Easy provides students with a convenient way to exchange their old belongings, ranging from textbooks and electronics to clothing and furniture.

The platform offers a user-friendly interface where students can create listings for items they no longer need, set their own prices, and connect with potential buyers within the campus network. By leveraging the power of peer-to-peer transactions, Sell-Easy aims to reduce waste, promote resourcefulness, and foster a sense of community among students.

Automated Free Slot Finder

Thu, 27 Jan 2022 00:00:00 +0000

The VIT Common Slot Finder is a project that employs computer vision to analyze complex student timetables and identify available free slots. Designed for internal VIT student teams, it facilitates meeting scheduling by analyzing the timetables of approximately 100 students to find common free slots. By uploading their timetables, the tool returns the common free slots for the respective day, ensuring efficient team coordination.

The sample timetable uses red marks to indicate students' class times. Our model analyzed all timetables, creating a JSON file of all empty slots. This data was then compared across students to identify common free slots. If there were any clashes, the model provided insights into other potential meeting times, indicating which students had classes during those slots. If a common free slot was available, it was displayed; otherwise, the next best time with the fewest students in class was shown.

Counsailia

Thu, 27 Jan 2022 00:00:00 +0000

Counsailia is an AI-driven counseling service designed to connect students with counselors who are their seniors or individuals who have walked a similar path. It starts with a comprehensive questionnaire that tracks students' interests and assesses their current standing. Our algorithm then analyzes this data to create tailored pathways for their interests, allowing students to explore specific fields. For additional support, we connect students with experienced senior peers, providing personalized guidance.

These senior students received a fee for their counseling services. Especially post-pandemic, when many students struggled to connect with seniors due to online education, Counsailia has been instrumental in bridging this gap. Developed with the goal of helping students choose the right path, Counsailia empowers students by providing targeted advice and mentorship.

Data Pre-Proccessing CLI

Wed, 27 Apr 2016 00:00:00 +0000

The CLI was developed for efficient data preprocessing, crucial in refining raw data for analytical tasks. Through a user-friendly command-line tool, it streamlines tasks such as cleaning, encoding, and visualization, thereby empowering data scientists to optimize workflows and ensure data integrity. Key functionalities include handling missing data, resolving inconsistencies, and selecting relevant features, all aimed at enhancing data quality and improving machine learning model performance. The framework utilizes a customer dataset from a credit card company, focusing on preprocessing techniques tailored for predictive modeling.

By addressing challenges like class imbalance and inconsistencies within variables like salary, the project employs command-line tools for data manipulation and resampling. The seamless integration of preprocessing steps into the modeling pipeline enhances reproducibility, automation, and scalability, facilitating the development of robust models capable of predicting customer purchase behavior effectively.

Linked Business S.A.

Wed, 27 Apr 2016 00:00:00 +0000