Pavithra Ezhilarasan

College Park, MD ***** +1-240-***-****

ac4xwg@r.postjobfree.com http://pavithraezhil.weebly.com https://www.linkedin.com/in/pavithraezhilarasan

EDUCATION

University of Maryland –College Park, MD Expected May 2018 Master of Science, Telecommunications Engineering

Coursework: Machine Learning, Internet of Things, Cloud Computing, Networks and Protocol PES Institute of Technology -Bangalore, India June 2015 Bachelor of Engineering, Electronics & Communication Engineering Coursework: Probability and random process, Calculus, Signal processing Coursera certifications: Algorithms, Python for Data Science, Cloud Computing

WORK EXPERIENCE

Paradyme Management, Greenbelt, MD

Corporate Technical Intern September 2017- Present Technology stack: Python, Google NLP, NLTK

• Enhanced parsing for the Resume Skill Repository project using Pytesseract, word-embeddings and Regular Expression

• Performed detailed debugging and testing of the resume skill tool and documented the error report PES Institute of Technology, Bangalore, India

Research Associate (Professor V.Ramasubramanian), Text-To-Speech Laboratory August 2015-July 2016 Research Paper: Mixed-Viterbi unit-selection and variants for syllabic TTS with optimal monophone back-off Technology stack: MATLAB, Festival, C programming

• Enhanced speech recognition algorithm for Text-To-Speech system for Kannada, an Indian language

• Developed a variant of mixed Viterbi algorithm in MATLAB using monophones and syllables for speech decoding

SKILLS

• Programming: Python, C++, SQL, Java, MATLAB

• DevOps: Windows, Linux, Amazon Web Services (AWS)

• Libraries/Framework: TensorFlow, Weka, Spark MLLib, NLTK, Git, Cassandra, Tableau, Apache Spark, Kafka, OpenCV

• Machine Learning: Bayes, K-NN, PCA, LDA, K-means, Perceptron, SVM, Neural Networks, CNN, LSTM-RNN, HMM

PROJECTS

Automatic Photo caption generator December 2017

Technology Stack: Python, Keras, TensorFlow, AWS (EC2)

• Used Flickr8k photo dataset to fit a pre-trained VGG model for feature extraction and fit a CNN model

• Implemented the merge model of RNN for image captioning and used BLEU scores for testing accuracy Motion detection from IP camera feed using Kafka, Apache Spark pipeline March 2018 Technology Stack: Apache Spark, Kafka, Zookeeper, Maven, Java, HDFS, OpenCV

• Processed video content from a cluster of IP cameras using Zookeeper and OpenCV

• Deployed a scalable, fault-tolerant system for motion detection using Kafka, Apache Spark and used HDFS for storage Face recognition using Machine Learning Classifiers March 2017 Technology Stack: MATLAB

• Implemented Naïve Bayes and K-Nearest Neighbour algorithm using Liner Binary Pattern features on dataset with pose, illumination variation

• Implemented dimensionality reduction algorithms: Principal Component Analysis(PCA), Linear Discriminant Analysis(LDA) which decreased computation time and further increased the accuracy by 45% Twitter Sentiment Analysis using Twitter API in Python May 2017 Technology Stack: Python, Twitter API

• Crawled tweets specific to the subject of interest using tweepy and tagged them according to POS (Part of speech)

• Gauged the sentiment of tweets by classifying them as positive, negative or neutral using a Naïve Bayes Classifier

ACTIVITIES

Director of Public Relations, American Society for Engineers of Indian Origin (ASEI) August 2017-Present

Contact this candidate