Github LinkedIn Website
Machine Learning Intern Synopsys May 2017- Aug 2017 Unsupervised Anomaly Detection
• Developed an anomaly detection system for the company’s expense data using Machine Learning.
• Used Kernel Density Estimation and a one-class Support Vector Machine in a hybrid approach that led to achieving an F1 score of 0.87.
• The internal audit team deployed the system instead of manual data examination
• Other project - Developed python programs to keep track of Cost Center and G/L Accounts mapping changes for BW data conversion to SAP S/4 HANA system
Research Assistant University of Massachusetts Jan 2017 – May 2017
• Built an automatic dietary monitoring system which classified incoming audio streams of mastication from the Bluetooth headset worn by the user into different food categories.
• Developed a semi-supervised approach with Deep Boltzmann Machine for unsupervised pre-training and a deep feedforward neural network for fine-tuned classification. We achieved an accuracy of 94% with the DBM-DNN model which outperformed previous systems at the time.
Lowell, MA UMass Lowell Aug 2016 – May 2018*
• M.S in Computer Science, May 2018*. GPA: 3.5
• Graduate Coursework: Algorithms; Artificial Intelligence; Machine Learning; NLP; Big Data System Design; Databases; Human Computer Interaction; Computer and Network Security. Mysore, India VTU Aug 2012 – May 2016
• B.E in Computer Science, May 2016. GPA: 3.6
• Undergraduate Coursework: Operating Systems; Databases; Algorithms; Programming Languages;Comp. Architecture; Engineer- ing Entrepreneurship.
• Memory Augmented Chatbot (present). Working on a neural conversational model that distills knowledge from external facts into a working hierarchical external memory to augment its capability to produce coherent and contextually relevant responses in a conversational setting. Pytorch, Python
• Linguistic Diagnostic Toolkit (2017 - present). An NLP toolkit that investigates the linguistic relationships captured by different word embedding models through intrinsic evaluation and also evaluate them on a set of common NLP downstream tasks such as NER, POS Tagging, SRL, Sentiment classification, Relation Extraction etc using neural architectures. Python, Chainer
• Semi supervised Learning with Deep Convolutional Generative Adversarial Networks (2017). Aims to leverage the representations learnt by the discriminator network during the adversarial training process and use them to achieve one/few shot learning in supervised CNNs in the task of image classification using MNIST and CIFAR 10/100 datasets. Python, Keras, Tensorflow
• Clinical Concept Classification using Bidirectional Long Short Term Memory Network(LSTM) (2016). Clinical concept classification system for medical notes in the 2010 I2B2/VA challenge dataset using Bidirectional LSTM with GloVe word embeddings trained on the MIMIC II dataset. Python, Tensorlfow, Keras
• In-one file manager (2015-2016). Desktop application that logically aggregates files based on their file extension. It helps to store information regarding file organization in storage media like DVDs, HDDs etc. It reminds the user to backup important documents and photos. It suggests the user to rename vaguely named documents like PDF files, word files etc. Java, JavaFX, SQLite Additional Experience and Awards
• Second Prize, CS Dept Project Expo: Awarded 2nd prize for In-one file manager, out of 35 projects.
• Deep Learning, a 5-course specialization by deeplearning.ai on Coursera. Specialization Certificate earned on February 3, 2018 Skills
• Languages : C++; C; Java; Python; SQL; PHP
• Libraries : Tensorflow; Chainer; OpenCV; Torch; Scikit-learn; Numpy; NLTK; CoreNLP; Pandas; SQL; MongoDB; MySQL;
• Technologies : Visual Studio Code; Eclipse; MATLAB; Atom; Pycharm; Git; Tableau; Docker Areas of Interest : Software Development; Machine Learning; Artificial Intelligence; Data Science; Computer Vision; NLP