• Developed a solution for predicting the stability of coating during the process using time-series-based machine learning models• Designed a model architecture for stability prediction by fusing time-series based sensor data with on-process visual data of coating process using feature vectors• Implemented variational autoencoder (VAE) and segmentation models for feature extraction of anomalies from visual data during coating process• Created a preprocessing algorithm for visual… Show more • Developed a solution for predicting the stability of coating during the process using time-series-based machine learning models• Designed a model architecture for stability prediction by fusing time-series based sensor data with on-process visual data of coating process using feature vectors• Implemented variational autoencoder (VAE) and segmentation models for feature extraction of anomalies from visual data during coating process• Created a preprocessing algorithm for visual data using ellipse detection to handle variations in zoom, brightness, and noise; enhanced feature extraction by 15%• Developed a predictive model for stability prediction using LightGBM (LGBM) and Long Short-Term Memory (LSTM) models with fused sensor & image processed data• Built a solution for post coating process to compute properties and evaluate pass/fail criteria using computer vision• Designed a model pipeline for property evaluation using a segmentation model followed by image processing techniques to compute the property features of the segmented layer using TensorFlow and OpenCV libraries• Utilized image processing techniques, including filters, GrabCut, and Gray Level Co-occurrence Matrix (GLCM), to develop solutions for coating property evaluation• Worked on the backend development of a coating evaluation web application for customers, including REST APIs, database management, and containerization Show less

• Developed an AI based vision system that assist the radiologist with identifying and localizing the abnormalities in the chest using X-ray images• Achieved a 24th Rank out of 1300 participation in the VinBig Kaggle competition

• Built an automated system that enables non-specialists to perform ML/DL tasks using a GUI interface• Designed common training and inference workflow of the state-of-the-art deep learning models for classification, object detection, and segmenation predictive tasks

• Modulus Housing invented foldable cabins to solve construction industry problems.• Designed and prototyped the double bearing telescopic sliding mechanisms for walls and beams of the cabins