I am a Computer Science Graduate at UT Dallas with a professional experience of 3+ years in developing Big Data and ML Applications.
Actively working in the areas of Big Data Applications with Deep Learning, Machine Learning & Natural Language Processing, Software Development and Web Applications.
August 2016 – December 2018
Coursework: Machine Learning, Natural Language Processing, Big Data Analytics, Database Design, Web Programming Languages, Implementation of Advanced Data Structures and Algorithms, Design and Analysis of Algorithms, Statistical Methods for Data Sc.
August 2011 – May 2016
Coursework: Design and Analysis of Algorithms, Data Structures, Operating System, Database Management Systems, Object Oriented Programming
January 2019 – Present
- Productionalized and scaled personalized recommended systems in the real estate domain for 16M users on a daily basis and improved the user engagement by 6% and reduced the performance time by 20% using PySpark, AWS Athena and Elastic MapReduce
- Lead, designed and deployed the architecture of large scale machine learning data pipelines based on terabytes of user behavior data integrated with AWS Glue for ETL transformations, Sagemaker for hosting ML Inference endpoints orchestrated with Step functions
- Developed a platform to track the performance of the ML models to analyze the business impact user engagement, and retention
- Developed the personalized notifications and email campaigns for new listings in the market, recommended listings based on the user’s activity and A/B tested with multiple model variants for different audience segments
- Research, design and implemented ML modules on the website to improve buyer experience like Similar Homes to the home the user is viewing, Affordability Calculator which recommends affordable neighborhoods
- Developed Real profile, which gives the portfolio of the user based on his activity in the past and deployed it as a candidate generation model. Developed a robust model based on LightGBM, Match Score, which gives the relative importance of any listing for a user and deployed it as a candidate ranking model for multiple use cases
January 2018 – August 2018
- Designed a framework to create state chart XMLs. Developed the front-end Graphical User Interface using Angular 4
- Developed the back-end of the framework using Python, which builds the SCXMLs from the flow charts on the GUI
September 2017 – December 2017
- Developed the accounts application for sign up and sign in of a user and registered organization on the VOMO platform using Node.js and deployed it on AWS EC2
- Analyzed data to recommend volunteer projects for users using machine learning models, content based and collaborative recommender systems
May 2017 – August 2017
- Developed integrated framework and sample application using Apache SOLR, Hadoop and Spark for improving the quality of big data solutions on large scale datasets.
- Optimized the indexing and search performance of Apache SOLR using Java, Spark and Web Services built on Spring MVC in Cloudera environment which uses Hadoop Distributed File Systems(Cloudera).
November 2016 – January 2017
- Designed web pages of Prof. Yvo Desmedt using HTML, CSS, JavaScript and exposure to tools like Latex.
Hadoop, Spark, Kafka, SOLR, Elastic Search, Kibana, Scala, TensorFlow , AWS EMR, Glue, Sagemaker, SQS
Spring MVC, REST Web Services, Angular 4, Node.js, HTML, CSS, JavaScript, jQuery
SQL (AWS Athena, MySQL), NoSQL (MongoDB, DynamoDB)
Git, Shell, Bash, UNIX/Linux, Excel, Agile
- Implemented BFS, Dijkstra, A* algorithms on real-time google maps data to compute the shortest distance between two locations and extended it by implementing Euler tours on these maps.
- Implemented Multi-dimensional search analogous to amazon selling tens of thousands of products for adding, removing and fetching items of similar prices and similar descriptions and generating invoices using TreeMap, HashMap and HashSets.
- Implemented Big Integers and arithmetic operations on Big Integers in Java, various versions of Merge Sort (avoid copying, combining with insertion sort) and Quick Sort (Dual Partition, Random Pivot Partition) on Generic arrays.
- Implemented an application analogous to real-life banking application using Spring Boot framework. Developed the front-end using Angular 4 and the backend REST Services using Java. Integrated MySQL using JPA.
- Electronic Bidding Website developed for both buyers and sellers, where products are sold by auctions. Developed the back-end functionality using REST services, Spring MVC and Hibernate. The front-end is developed using HTML, CSS, JavaScript, JQuery.
- Implemented a messenger application where users can exchange messages using MEAN stack. Developed the front-end using Angular 2 and backend using Node.js. Integrated MongoDB to store users and messages.
- Predicted the sentiments of tweets using nltk library in Spark environment by streaming the tweets over a messaging server, Kafka and visualized them over a global map by using Kibana where tweet objects are indexed using Elastic Search.
- Encoded real time news feeds to event types using SVM and multinomial Naive Bayes in Spark environment where feeds are queued over Kafka server. Text analysis on news feeds is performed using CoreNLP and SEMAFOR and knowledge base, WordNet.
- Identifying the author of an arbitrary e-mail by using customized writing style features like lexical, syntactic, idiosyncratic attributes and syntactic n-grams using Stanford coreNLP and scikit-learn and by applying Support Vector Machines, Random Forest, Neural Networks and Boosting Classifiers.
- Predict which Bosch parts fail quality control along the production lines using Ensemble methods using boosting, bagging and random forest classifiers using Python. (Competition hosted on Kaggle.com)
- Topic Modelling of the presidential debates to detect the main agenda of the candidates using MALLET. Detecting the audience reactions and sentiments by interventions using NLTK.