Projects

  • Face Mask Detection

    Built a machine learning model to detect face masks in images having full-frontal faces, by performing feature extraction using pretrained CNN models like ResNet50. Performed classification of features using decision trees, logistic regression, and SVM model. [Python 3, TensorFlow, scikit-learn]

  • Text Classification, Prediction and Bias Extraction using NLP

    Developed a Naïve Bayes classifier with unigram bag-of-words features and a Logistic Regression classifier that identified hate speech in a real-world hate-speech classification dataset. Implemented word and character level n-gram language models and character-level RNN model to predict the next word given a sequence of words in a text. Used pre-trained GenSim models to extract similarities, analogies, semantics, and visualization of word embeddings to identify cultural biases in text. [Python 3, PyTorch, word2vec]

  • Detection of Traffic Sign and Lights

    Designed and implemented a program that detected traffic signs and lights using Hough algorithms to identify shapes and select objects based on pixel location and properties.

  • Patient Nutrition Tool

    Built a web application for hospitals to enable customized nutrition and diet restrictions of patients based on their medical history of allergies and medication. [Java Spring Boot, MyBatis, MySQL, FHIR server]

  • A Visualization Approach of Business Location Recommendation

    Developed an application that allowed potential business owners to choose the geographic location of new businesses. Various approaches like sentiment analysis(NLTK), LDA topic modeling and K-means clustering(scikit-learn) were used to study and visualize business data from the Yelp dataset. [Python 3, React, Tableau, NLTK, scikit-learn]

  • Machine Learning for Trading

    Created a portfolio management system for trading using different machine learning algorithms like Decision Trees, Random Tree and Ensemble learners to beat the benchmark prices. [Python 3, Numpy, Pandas]

  • Solving Raven’s Progressive Matrices

    Designed an AI agent which solved 2x2 and 3x3 Raven’s Progressive Matrices problems. The agent used Knowledge-Based AI concepts like case-based reasoning, semantic networks, and production systems to determine the optimal solution from the images and solve the problems. [Python 3, Pillow]

  • Artistic Strokes for Images

    Built an application in Python 3 and OpenCV that would render photographs in various painting style filters. The final image is formed by merging multiple layers of painterly layers. Each layer represents an image rendered with brush strokes of a specific radius, using different style parameters to simulate strokes in various styles like Impressionist, Expressionist, Colorist Wash, Pointillist or a Generic style. [Python 3, OpenCV]