Slack Bot for GitHub users

Developing a Slack bot that discovers relationships between source codes and developers by mining GitHub repositories. ComBot uses repository contributions for creating file-to-developer mappings & facilitates quick communication with the developers via Slack.

Note: Ongoing project in a private repository. Source code will be made available soon.

Meeting Room Scheduler

Implemented a proof-of-concept prototype for scheduling meeting rooms at an organization at Cloudera’s intern hackathon. Features included waitlisting, reshuffling on cancellations, priority scheduling, etc.

Note: This project involved the use of some of Cloudera's proprietary code. Hence it cannot be made publicly available.

Twitter Sentiment Analysis using Spark Streaming & Kafka

Designed and developed a Spark Streaming application integrated with Apache Kafka to perform real-time sentiment analysis of incoming product review tweets.

Performance Optimization of SimpleDB

Made contributions to the open-source database system ’SimpleDB’ (Boston College) in tasks such as reduction of buffer look-up time, and cache hit optimization by implementing a ’Least Recently Modified’ buffer replacement policy in Java.

Library Management System

Designed an ER model for a university library database system, translated it into relational model using Oracle 11g and enforced the various key and integrity constraints imposed by requirements.

Behavioural Targeting using Web Activity & Social Media Data

• Devised a hybrid approach which involved monitoring a user's web activity and extracting social media data to create a user profile for behavioural targeting.
• Built a fine-tuned unbiased SVM classifier using linear kernel with an accuracy of ~88.5% to segment users into one of the 6 predefined categories based on their areas of interest.
• Developed a dynamic word cloud module to visualize the top keywords from user's browsing history.
• Technologies used: R, Python and MongoDB.

Music Recommender System

Developed a recommendation system based on Alternating Least Squares Method for Collaborative Filtering using Apache Spark’s scalable Machine Learning library (MLlib). Performed parameter tuning for building the model and used RDD caching mechanism to optimize the execution time.

Note: This project was developed for an academic course. The code has been committed to NCSU's GitHub domain account.