M.S. Computer & Information Science
University of Delaware, Newark, DE.
B.S. Computer Engineering
University of Pune, India.
Presently, I am working on Energy aware scheduler for Hadoop which makes use of the heuristic algorithms to minimize the energy consumption of the MapReduce applications.
As a part of XSTACK DynAx project, I worked on enabling inter-operatibility of Exa-Scale Framework – SWARM with that of legacy code in MPI. Developed benchmarks to demonstrate MPI+SWARM inter-operatibility on the shared memory system and also performed comparative study of the results.
NASA funded project targeting at detecting extreme events, hotspots of change and clusters in the plant productivity data from remote sensing MODIS satellite.
Pattern analysis, model formulation, pattern prediction on the climate data from DAYMET using clustering algorithms. Statistical analysis to find correlation between climate and plant productivity data.
Graded labs, assignments, held lab sessions and office hours for students.
- CPEG323 : Introduction to Computer Systems Engineering
- CPEG324 : Computer Systems Design I
- CISC 106 : General Computer Science for Engineers.
Jr. Spftware Engineer
Worked on development of web application – ‘sawangadi’ for commodity traders and farmers.
Designed database architecture and PHP framework.
- Java Script
- Map Reduce
Libraries and Utilities
- Visual Studio
Energy Aware Hadoop Scheduler
Developing EnergyAware scheduler for Hadoop’s YARN to minimize energy consumption of MapReduce application.
Exa-Scale SWARM MPI Interoperatibility
Developed various benchmarks to demonstrate MPI+SWARM inter-operatibility. Worked on the technical memo which showed the advantages of this approach.
NASA Carbon Monitoring and Upscaling Framework
Worked on pattern analysis, model formulation, pattern prediction on climate data from DAYMET.
Statistical analysis to find correlation between climate and plant productivity data.
Scalable Many Core Memory Consistency Model Study
Performed literature survey of state of art techniques in NoC, coherency, memory consistency.
Developed Theoretical model which can scale up to 256 cores.
Vector Space Model for Feature Location
Implemented technique to convert code and queries to VSM. Which returns results for the queries entered by users. Evaluation show results better than tradition feature location techniques.
Medical Data Analysis using Spark on IBM BlueMix Containers
Developing algorithms to extract interesting pattern in medical datasets using Spark inside Docker containers. Performance will be tested on IBM’s Bluemix platform.
Fraud Detection System Using Hadoop
Implemented Markov chains for learning and fraud detection using Hadoop.
Tested the model on the synthetic dataset.
Measurement of Memory Hierarchy Parameters
Developed benchmarks to measures parameters of cache like latency, capacity, associativity for various levels of cache. Evaluated results on three platforms for correctness.
Parallelize NU-MineBench Benchmark
Parallelized sequential implementation of Kmeans using openMP and OpenACC.
Achieved speedup of about 26 times in OpenACC implementation.
Developed a web based framework to emulate GUI that enabled access to a remote Linux server using web browser to emulate Operating system functions.
Legacy MPI Codes and its interoperability with fine grain task-parallel runtime systems for ExascaleETi TechMemo
Vulnerability of Western North America to Extreme Climate EventsPoster
A Study of Memory Consistency ModelReport
Volunteerism, Services and Awards
Paper Sub reviewer
IPDPS’14, DFM’15, CGO’16, ESPM2’16, ASPLOS’16
Excelsior Technical Event, University of Pune
Indian Graduate Student Association, University of Delaware.