@lr & Email:
& Phone: +91 879 385 8440
@cccc Year & Degree / Certificate & University / School & CPI / %
2019 (expected) & M. Tech (Computer Science and Engineering) & IIT Bombay & 8.75 / 10
2015 & B. E (Computer Science and Engineering) & Dr. BAM University, Aurangabad & 68 %
2011 & Class XII (Maharashtra State Board) & Deogiri College, Aurangabad& 70 %
2009 & Class X (Maharashtra State Board) & Holy Cross High School, Aurangabad& 90.30 %
Machine Learning, Natural Language Processing, System Administration, Software Development.
System Administrator
Dept. of CSE, IIT Bombay, Mumbai (July 2016 - Present)
Responsible for maintaining, upgrading and monitoring different software services such as e-mail, DNS, LDAP etc. used by the department
Configuration management of more than 400 lab systems using Puppet and Ansible
Implemented VLAN for CSE Department
Supervised the installation and maintenance of Biometric-based entry system
Fake news Detection (M.Tech. Project)
Guide: Prof. S. Sudarshan (IIT Bombay), Co-guide: Prof S. Chakrabarti(IIT Bombay) (May 2018 - Present)
Objective: To create an API for identifying the fake text articles and images
Designed and implemented custom scrapers using Scrapy framework to get structured data from from various News sources
Enhanced the Image+Text matching algorithm to identify images used out of (original) context
Used Social media analysis for estimation of source credibility
Future Scope: Improve source credibility estimates for Indian News and media sources
Tools Used: Python,Scrapy,Flask,Flutter,IBM watson API,Apache Solr,Elasticsearch,nltk
Studying methods for identifying fake news. (M. Tech. Seminar)
Guide: Prof. S. Sudarshan (IIT Bombay) (Jan-April 2018 )
Studied various methods used for Fact checking from different types of Knowledge sources
Studied methods used to estimate source credibility and verification
Memory Augmented Neural Machine Translation (Research and Development Project)
Guide: Prof. Pushpak Bhattacharya (IIT Bombay)(July 2017 - December 2017)
Solved the problem of LSTM based NMT models drowning out the signals of infrequent words in corpus
Implemented Memory-Augmented NMT model with separate memory elements for infrequent words
Network Task manager for Linux OS (B. E. Project)
Guide: Prof. Madhuri Joshi (June 2014 - April 2015)
Designed and developed a Linux application for tracking network processes using Python and PyQT as front-end
Used GeoIP and Google Maps API for plotting the geographical path a packet takes to reach a remote server on Google maps
API for Inter-VM communication using shared memory. (CS695: Cloud and Virtualization , October 2016)
Objective: Enable communication between VMs using shared memory.
Built a wrapper API over IVSHMEM library to provide a shared memory interface for communication between a guest OS and the host
Human activity recognition using Smartphone (CS725: Foundations of Machine Learning, April 2017)
Objective: To identify user activity using data from various sensors in a Smartphone
Modelled the problem as a multi class classification problem. Used various classifiers like Gradient boosting ,k-nearest neighbour, Random forests etc to get maximum accuracy.
Sentiment tracking across time (CS635: Web Search and Mining, November 2017)
Objective: To track sentiment across time for a web source and identify temporal patterns
Used NLTK to generate sentiment scores and performed a time series analysis on generated scores across all articles in a web source
Keystroke detection using keyboard acoustic signals (CS 753: Automatic Speech Recognition, November 2017)
Objective: To recognize a keystroke using its acoustic signals
Generated Data which mapped audio to keys on keyboard using custom scripts written in Python
Extracted MFCC feature vectors of the individual keystroke audio and trained an SVM classifier with a dictionary and language model on it with an accuracy of 87% .
Optimized Keyvalue store (CS 744: Design and Engineering of Computing Systems, November 2017)
Objective: To construct a basic key value store and improve its performance by applying different system level optimizations
Analyzed the characteristics at peak utilization and did extensive profiling using Valgrind.
Applied various optimizations to improve the performance of the system by 200% compared to baseline system
Headline generation using text summarization (CS 726: Advanced Machine learning, April 2018)
Objective: To generate headlines given an article using various deep learning methods
Modelled this problem as monolingual machine translation task . Applied various machine translation techniques like SMT, NMT and variants to generate semantically accurate and succinct summaries
Studied and implemented various text summarization algorithms based on RNN, CNN and Pointer Generator networks
Class Representative (2016-18)
Elected unanimously by the batch, represented batch in academic and CSEA council
Organized various department level sports, cultural and academic events, acted as first point of contact between faculty and students
Interview Coordinator (2016)
Assisted in the placement of 1600 students within a team of 200 students over a period of 16 days
Was appointed as Interview Coordinator for A.T Kearney and Works Applications
Department Placement coordinator(2018-present)
Organized and conducted regular coding and aptitudes tests to prepare batch for placements
Coordinated with institute placement team to ensure smooth functioning of the placement process
.35
Foundations of Machine Learning
Organization for Web information
.32
Automatic Speech Recognition
Web Search and Mining
.32
Cloud and Virtualization
Advanced Machine Learning
[]
Placed in top 0.7 percentile in GATE-CS (March 2016)
Ranked 2nd in state in Graduate Excellence Examination (March 2014)
Placed 2nd in University level project competition ’Proyecto’ (April 2013)
Maharashtra State talent search scholar (2008)