I have worked on several projects in the field of Database Systems, Distributed Systems,Computer Vision and Image Processing, Machine Learning, and Data Intensive Computing during my course of study at University at Buffalo. Below is the list of all the projects that I have worked on so far.
It is a simple SQL query evaluator with support for Insert, Delete, Select, Nested-Select, Project, Join, Bag Union, Limit, Aggregate functions (COUNT, MIN, MAX, AVG, SUM), GROUP BY,
GROUP BY Aggregates and ORDER BY clause on Big Data (TPCH). It also supports Indexing to some extent.
It is a Dynamo-style key-value storage implementing partitioning, replication and failure handling
to provide per-key linearizability and availability even under failure.
A Big Data pipeline to perform Data Cleaning and then word count and word co-occurrence algorithms on the text data collected from Twitter REST API, New York Times API, and Common Crawl Data on Sports and then performed visualization in Tableau.
To analyze Influenza outbreak by performing EDA on the Twitter data extracted by using Twitter REST APIs and comparing the data with Official Influenza Statistics using charts and heat map.
A distributed hash-table based on Chord that provides node joins, ID space partitioning, and ring based routing.
A group messaging android application with decentralized TOTAL and FIFO message ordering guarantees.
Implemented hough transform algorithm to detect lines, vertical and diagonal, and circles in the image.
Implemented template matching algorithm to find the template in the given image, invariant of template size.
Implemented K-means clustering algorithm to select the color palette for color quantization of an image.
Implemented GMM using Expectation Maximization Algorithm on Old Faithful Dataset.
Implemented morphological operations, Opening, Closing, Dilation and Erosion and then using them to remove noise from an image and extract boundaries.
A basic study on the effects of various hyper-parameters on a neural network model used for the task of classification.
Using Deep Reinforcement Learning Algorithm – Deep Q-Network to teach the agent to navigate in the grid world environment in order to reach the goal.
Using Logistic regression, Neural Network, Random Forest and SVM on the MNIST and USPS Dataset. Further, implemented ensemble of these four classifiers using Majority Voting.
Implemented both the closed form solution and Gradient Descent solution for linear regression on the LeToR Dataset released by Microsoft Asia.
Implemented linear regression, logistic regression and Neural Network on Human Observed Features Dataset and GSC Features Dataset extracted from CEDAR Letter Dataset which consists the image snippets of the word “AND” to compare handwriting.
Using Sobel Operator to detect horizontal and vertical edges in an image.
Implemented first three steps of SIFT in order to detect keypoints in an image.
Warping two images using the Homography matrix computed with RANSAC.
Using thresholding to separate object in foreground from the background.
Detecting porosity in an image using point detection algorithm.
Android application which enables two android devices to send messages to each other.
To understand the concept of epipolar geometry which includes computing the Fundamental matrix with RANSAC, selecting inlier match pair and for each keypoint in the left image draw epilines on the right image and vice-versa and then computing disparity map of two images.