Projects
Osteoarthritis Screening with Deep Learning
Conducted comprehensive Exploratory Data Analysis (EDA) on diverse medical X-Ray images, utilizing Python and implementing a TensorFlow algorithm for automatic hand joint detection. Engineered a 6-layer deep learning model, achieving a 20% precision improvement in hand joint identification. Orchestrated model optimization, resulting in a 15% overall accuracy increase, surpassing industry benchmarks in medical image analysis with an MSE of 0.005 and MAE of 0.02.
View More
EcoScan
Engineered a responsive front-end with HTML, CSS, and JavaScript for EcoScan, featuring user-friendly Disease Detection pages. Collaborated on seamless integration with backend ML models for real-time plant health analysis.
View More
Portfolio
Engineered a standout portfolio site by leveraging HTML, CSS, and JavaScript. The site boasts captivating hover effects and is powered by the Locomotive Scroll library, ensuring seamless navigation. Applied responsive design principles for optimal cross-device performance, and integrated JavaScript to facilitate dynamic content loading and smooth scrolling, ultimately enhancing the overall user experience.
View More
Classification of Dogs and Cats Using Inception v3 Models
Developed and implemented two PyTorch-based models for image classification on the Kaggle Cats and Dogs Dataset. The SimpleCNN model demonstrated a validation accuracy of 85.24% and a loss of 0.4134, while the PyTorch Inception v3 model, utilizing a pre-trained architecture, showcased exceptional performance with a remarkable 94.32% validation accuracy and minimal loss of 0.0350. These models contribute to efficient and accurate classification of dogs and cats images, showcasing the versatility of PyTorch in deep learning applications.
View More
Credit Risk Assessment Model
Developed and deployed a credit risk assessment model utilizing Logistic Regression and Random Forest algorithms, achieving a notable 15% reduction in default rates while sustaining a commendable loan approval rate of 90%. Additionally, enhanced the credit risk assessment process by optimizing the model's threshold, resulting in a 20% reduction in false positives and minimizing potential losses.
View More
Task Management Application
Crafted a sophisticated and responsive task manager using HTML, CSS, and JavaScript, orchestrating seamless task management with features such as task addition, editing, and deletion. Demonstrated advanced proficiency in UI/UX design to ensure an elegant and real-time interactive experience for users. The resulting application reflects a combination of technical skill and thoughtful design for effective task organization and user engagement.
View More
