Computer Science State University

Mythili Sivakumar

480-***-**** ● **********@*****.*** ● linkedin.com/in/Mythili-Sivakumar ● github.com/Mythili895

Education

Arizona State University –Tempe, Arizona, United States May 2018 (expected)

• Master’s in computer science, CGPA: 3.74/4.0

Amrita University – Coimbatore, India August 2012-May 2016

• B.Tech Computer Science, CGPA: 8.6/10.0

Skills

Languages: Python, MySQL, Java, C, C++, MATLAB, SQLite Web Application Technologies: HTML, CSS, JavaScript, FLASK, REST, Bootstrap, AJAX Tools and Libraries: Numpy, Scikit, Pandas, NLTK, R/R Studio Big Data Tools: Cassandra, Hadoop, Spark

Research Experience

Actionable Analytics Lab, Arizona State University Dec 2017 – Current Graduate Student Assistant

• Research Objective: To improve the efficiency of a web crawler using Machine Learning Techniques. Amrita Multidimensional Data Analytics lab, Amrita University Aug 2014 – Jun 2016 Undergraduate Research Intern

• Research – “A Novel Approach for Aspect Identification using Semantic Networks”

• Identified aspects of a given search query to diversify search results using Semantic Networks. Proposed a greedy heuristic over the Louvain modularity algorithm that improved the precision by 4% and recall by 2%.

• The precision values were better than the existing semantic search engines yippy.com(Clusty) and sensebot.com. Achieved an accuracy of 68%

• Visualised the semantic network using Gephi, a graph visualization tool. Projects

Adaptive Quiz System

• Developed an intelligent tutoring system, a web application that provides java quiz sessions to the students based on their competency in Java.

• The system motivates the students by providing open social student modelling, where the user understands his levels compared to his peers. Used content-based recommendation algorithm and the systems adapts to the concept hierarchy and the difficulty of the questions.

• Language/Framework: FLASK, HTML, SQLite, D3.js, CSS, Python, JavaScript Twitter Stream Analysis using Spark and Cassandra

• Developed a cloud-based system to perform analytics on twitter stream data.

• Used Cassandra as distributed database to store the nearly real time streaming of twitter data due to its fast write capability and spark as an execution engine to perform in-memory analytics.

• Performed log management and monitoring of the distributed system using Sematext.

• Language/Framework: Spark, Cassandra, Sematext, Apache Lucene Classification of Facial Expression using Active Learning

• Identified facial expression classes on MMI dataset using active learning strategy.

• Performed comparative study between 2 different learning strategies: Uncertainty based sampling and random sampling.

• Language/Framework: MATLAB

Visual Question Answering Using Deep Learning

• Built a system that solves the challenge of answering open-ended question related to an image using neural network.

• With the phrases related to the image as input, the system uses Long short-term networks (RNN) to answer the questions.

• Language/Framework: Keras, Python

Amrita Deakin University Project

Epidemic Spread Prediction Using Crowdsourcing and Predictive Analytics

• Built an android application that is a crowd sourcing platform for reporting incidence of malaria.

• Used ARIMA model to detect and predict the spread of malaria based on factors like temperature, humidity and rainfall. Used Google Maps API and its Markers to visualize the incidence value.

• Language/Framework: Java, Android Studio, JRI

Spark API for Geo Spatial Operations on Hadoop Distributed File System ACM SIGSPATIAL CUP 2016 PROBLEM

• Applied spatial statistics to identify significant hotspots in New York City yellow cab taxi records using Getis-Ord statistic.

• Architecture: A distributed computing framework with 3 instances having Hadoop as the file system and spark as an execution engine. The cluster performance was monitored, and various aspects of it were studied using Ganglia.

• Language/Framework: Scala, SparkSQL, GeoSpark

Contact this candidate