Looking for full-time opportunities in the field of Machine Learning to leverage my skill set and background in research, coursework and professional experience. I’m proficient with machine learning frameworks like Scikit Learn and Keras. My current research includes Graph Representational Learning using Deep Learning techniques. I’m familiar with distributed frameworks like Apache Spark and Hadoop. I usually use the AWS cloud to run my workloads.

RESEARCH INTERESTS

• Representational Learning on Graphs and Images.
• Exploring the Interpretability of Deep Learning Models like Graph Embedding models, Language Embedding Models like BERT and Computer Vision models like Inception models, etc.,
• Generative Adversarial Networks and its applications.
• Fake news identification and prevention using Machine Learning.

TECHNICAL SKILLS

• Python, Java, JavaScript
• Scikit-Learn, TensorFlow (Keras), spaCy, Apache Spark and Hadoop.
• MYSQL, MongoDB.
• HTML 5, CSS 3, Angular 2, React JS, Node JS.

CERTIFICATIONS

• Graduate Certificate in Big and Smart Data - Wright State University

ACADEMIA EXPERIENCE

Research Assistant - Wright State University. February 2019 – Present

Temporal Sentiment analysis during Pandemics.

• WORK IN PROGRESS

Vandalism detection in Wikipedia using user edit history.

• Work avaiable online at source code
• Collection of the Wikipedia articles and user edit history using Media Wiki API.
• Application of text vectorization and feature engineering techniques.
• Exploring Graph Representational Learning Algorithms like GraphSAGE and Node2Vec.
• Working towards replacing the bots (regex) currently in use by Wikipedia.

Nomophobia analysis from tweets using text classification models in spaCy.

• Work available online at source code
• Annotation of tweets collected using twitter API.
• Exploratory Data and visualization using Matplotlib.
• Pre-processing and vectorization techniques like Word Vectors, TF-IDF features are used.
• Experiments with various classification models like regression, SVM and naïve Bayes.

Teaching Assistant for CS 1160 – Intro to Python Programming.

• Designing Labs for the course.
• Educate under-graduate students about code versioning systems like Git.

ACADEMIC PROJECTS

• Implementation of core Machine Learning algorithms like regression, clustering and dense neural networks from scratch. source code.
  • Interpretation of the results using Matplotlib.
• Developed a Deep Audio classifier using convolutional neural networks which processes low dimensional Log-Mel features of an audio sample and classifies the audio sample as a drone buzz with or without a payload. source code
  • CNN is implemented using TensorFlow’s Functional API (Keras).
  • Interpretation of results using activation maps and classification metrics is performed.

• Implementation of a parallel sorting algorithm in a distributed setting using Google’s Remote Procedure Calls. source code

• Building Abstract Data Types in C++.
  • Stacks, Queues, Binary Trees, Heaps, etc.,

• Development of scripts to access docker containers deployed inside AWS EC2 instance using boto3 and paramiko libraries.

• Implementation of clustering algorithms like K-means and Fuzzy C-Means using Hadoop Java API.

• Building a simple search engine in Python with learning to rank algorithms.

PROFESSIONAL EXPERIENCE

Associate Consultant – (Web developer and Data Engineer)

Capgemini Technology Services India Limited. August 2015 - July 2018.

• Developed analytic runtimes in Python for analyzing real-time streaming time-series data and published the results in a dashboard built using Angular 2.
  • Usage of frameworks like Apache Spark and Hadoop to handle huge data.

• Developed a data aggregation pipeline and a job scheduler using the Spring Boot framework and Apache Spark that can proactively monitor the health of turbines and minimize the operational losses due to equipment failure.

• Building, refactoring and migrating applications from legacy/on-premise data centers to the AWS Cloud.
  • Migrated 5 Full-Scale Applications with a user engagement rate of 800 user sessions per hour to AWS Cloud Infrastructure with minimal to no downtime.
• Developed end to end responsive web applications with containerized microservices as backend, using Docker and deployed in the AWS Cloud.