PROJECTS
PROJECTS
• Automated image segmentation of X-ray images using U-Net neural network architecture to detect pneumonia on chest X-ray.
• Performed data preprocessing to generate masks from bounding boxes and converted them to one-hot encodings.
• Trained and tested the Keras model to achieve a dice coefficient of 97% and mean-IOU of 95% on the validation set.
Developed a credit risk model to predict whether a customer would default a payment using Amex customer application dataset.
Trained and tuned Random forest, Decision Tree, Logistic Regression, Naive Bias classifier to arrive at a final robust model.
Achieved precision of 0.72 and recall of 0.67 on validation set and a leaderboard score of 58.6 %.
• Collaborated with a team of 4 to develop a tool for simulation of linear & nonlinear dynamics of marine and offshore structures.
• Contributed to the frequency domain part of the python based tool that works as a web app as well as on command line.
• Extensively tested and validated the code for different vessel cases and successfully debugged and documented the codebase.
• Implemented K-means algorithm for clustering of companies using their stock market dataset from Yahoo finance.
• Performed data preprocessing using scikit package to finally cluster companies based on their stock movements.
• Used Principal component analysis (PCA) to reduce dimensionality and plotted the clusters in 2D.