Beyond the EU: DORA and NIS 2 Directive's Global Impact
Shubhangi nov20
1. SHUBHANGI TANDON
315, Cedar Street∙ Santa Cruz, CA - 95060 ∙ (831) 346 9274 ∙ shtandon@ucsc.edu ∙ LinkedIn ∙ Github
EDUCATION
• University of California, Santa Cruz California, United States Sep 2016 - Present
Masters in Computer Science (Current CGPA: 3.8/4)
Coursework: Algorithms, Machine Learning, Graphical Models, Computational Models and Discourse, Data Mining,
Programming Languages, Conversational Agents, Probability, Deep Learning
• Delhi College of Engineering (University of Delhi), Delhi, India Jul 2009 -May 2013
B. Engg in Information Technology: Graduated First Class with Distinction (75.02%)
Computer Graphics, Operating Systems, Computer Arch, Databases, Software Engineering, Information Technology
TECHNICAL SKILLS
• Languages and Frameworks: Python, TensorFlow, Word2Vec, Scipy, scikit-learn, Numpy, Keras, NLTK, Gensim, Pandas, Numpy
R, Java 8, Scala, C, C++, SQL, DB2, Imapala, Javascript, GIT,SVN, Jira, Maven, Gradle , Linux
• Statistics and Machine learning: Probability Distributions, Expectation, Bayesian Theory, MLE, Graphical Models, Deep
Learning, CNNS , RNN, Regularization, Regression, Classification, Clustering, Regularization, Kernel Methods, HMMs, Chatbots
WORK EXPERIENCE
• Researcher, Natural Language and Dialogue Systems lab, UCSC
• Generating End to End Natural Language from Meaning Representation with Seq2Seq frameworks and dual RNN encoder.
• Built systems using synthetic data augmentation, Teacher-Forcing, Open loop input to produce stylistically varied outputs.
• Research and Development Intern, VMware, Palo Alto Jun 2017 – Sep 2017
• Cloud Management Business Unit: Built a subscription based notification system using Business Intelligence and predictive
modeling to help track the growth of an application in Beta phase for the Customer Success Team.
• Hybrid Cloud Admin Recommender (vSDA): A cloud admin that recommends how cloud resources (memory, disk, CPU)
should be scaled. Used a Decision Tree and implemented a cost minimization algorithm
• Software Developer, Goldman Sachs, Bengaluru Jun 2013 - May 2016
• Involved in designing, implementation, testing and maintenance of production applications. Took on multiple roles as
Technical owner, Technical Architect, Software Developer and Mentor.
• Developed an application that screens and reports written communication in the firm for regulatory violations on Java and
JavaScript based distributed services. Built the UI in Angular, used RESTful services for interfacing.
• Automated Batch Processes using Spring Batch for Sanctions System that screens firm data against Sanctioned Entities.
• Built a framework that captures and report Data inconsistencies as validation breaks to up-streams.
• Teaching Assistant at University of California, Santa Cruz Sep 2016 –Present
• Courses: Algorithms and Abstract Data Types, Introduction to Programming in Python
• Teaching topics like Time-Space complexity, Sorting, Searching, Dynamic Programming and Greedy Algorithms.
• Software Testing Intern for Windows Networking Division at Microsoft IDC Jun 2012-Jul 2012
• Constructed an automatic code generator tool (in C#) for a given UI Automation Framework
PROJECTS and PUBLICATIONS
• Discourse Coherence with Open Domain ChatBot (Link)
• Built a chatbot by enhancing a Bi-LSTM Seq2Seq Network with Multiplicative Attention using Tensorflow 1.1.0
• Added a dual CNN Encoder that injects context using Dialog Act Tag into the attention mechanism for coherent output
• Summarizing Dialogic Arguments from Social Media (Publication Accepted for SemDial 2017 Link):
• Classify dialogic data using a SVM Classifier to get 73% accuracy on Gun-Control Corpus
• Ensemble feature engineering and feature ablation from Stanford Core NLP, Penn Discourse Tree Bank, Rhetorical
Structure Unit, Speciteller , TextRank and LexRank.
• Topic based sentiment using Deep Nets (Link)
• Built a system for aspect level based sentiment analysis, using CNNS to generate custom joint word-topic-polarity
embedding and RNNs for sentiment prediction using the embedding for twitter data.
• Image Classification Models
• Transfer Learning on Stanford Clothing Attribute Dataset to classify the category of Clothing (e.g. Dress, Suit) using VGGNet
for feature extraction, and SVM, Logistic Regression for classification. (Github)
• Fashion MNIST Dataset: Predict the type of fashion image using Conv Nets in both Keras and Tensorflow. (Github)
• Sentiment Analysis on Yelp Dataset to predict the rating of reviews:
• Implemented Multinomial Logistic Regression, SVM and Naïve Bayes and built an Ensemble Python 78% accuracy
• Spam filter SVM, Logistic Regression, Naive Bayes on Enron Corpus to classify HAM/SPAM and compared their performance.
• Data mining on Iowa Housing and American Housing Survey Datasets
• Regression to predict the house of the price, Classification to classify its overall Quality,
• Clustering to cluster houses with similar features together and Apriori algorithm to infer association rules
• Natural Language Processing on Twitter Data
• Performed Latent Semantic Analysis to obtain compact feature sets, Lexical Normalization to enhance data quality
• Performed unsupervised learning using K-means clustering to group similar tweets.
• Applied Topic Modelling using Latent Dirichlet Allocation and Biterm Topic Modelling.