Enterprise Professional Networking Plateform
Worked on Enterprise Professional Networking Platform, where my accomplishments include:
- Setting up of the organization wide professional networking platform by leading the efforts to achieve necessary conformance with respect to security
- Aggregation and curation of data from multiple enterprise knowledge sources and transforming and storing it in a centralized schema for knowledge sharing
- Enhancements in the employee profile storage schema to facilitate high performance search
- Improvements in system performance by modifying the existing algorithms and the physical database schema
Big Data Testing Environments
- Engineered the CODD Metadata Processor for HP NonStop SQL/MX database, to allow the testing of query engine on big data by creating a metadata shell for big data, which is currently in use by HP for testing of its database engine
- Worked on simulating virtual hardware environments, to analyse the behavior of HP NonStop SQL/MX query engine for big data on futuristic hardware configurations
- Designed PCA based stochatistic data generation algorithm to generate synthetic database using the metadata statistics of an original database, such that the data distributions of the synthetic database are similar to the original database
Design and Implementation of PCM-Conscious “Group By” and “Having” Database Operators
PCM (Phase Change Memory) is a non-volatile memory which consumes very less idle power, has limited write endurance (106 to 108 writes per cell) and consumes high energy in writes. Reducing PCM writes is an important design goal of PCM-friendly algorithms. To achieve this goal, an hash table based algorithm was implemented in C for “Group By” and “Having” database operators, using small amount of DRAM along with PCM. It was followed by the simulations to verify the results using the PTLSim simulator.
Analysis of Inter-Modal Competition between High Speed Rail (HSR) and Conventional Transport Systems
High Speed Rail (HSR) is an eﬃcient mode of transportation for faster intercity movement. If HSR is introduced in a market, it’s going to aﬀect the equilibrium between existing transportation modes. The aim of this project was to understand the competition among diﬀerent modes of transportation, derive their optimal strategies (price and frequency) for proﬁt maximization and analyze the impact of introduction of a new transportation mode on existing modes of transportation using game theory.
Sphere Impostor using Tessellation Shader
The standard approach of displaying a sphere is to tessellate the sphere in smooth polygon mesh which leads to unacceptable polygon counts. Impostor is a technique where one uses approximation of an object up to rasterization. The raster pixels are then corrected to generate a pixel precise rendering of the object. This project was implemented using C and OpenGL 4.0, where the approximation of the sphere was created using the tessellation shaders, followed by the depth correction technique to display the sphere perspectively correct. Phong lighting model was used to calculate the lighting on each pixel.
Color Image Compression Using 2D-PCA and Normalization
2D-PCA is an image projection technique for image feature extraction. In this project, the 2D-PCA algorithm was implemented using MATLAB to store the color image in the form of feature matrices and principal components, which have very less size as compared to the actual image. After applying the 2D-PCA technique, normalization of the feature matrices was done to further reduce the size of the image.