Working on Ads Integrity

• Trained a machine learning model to classify cell roles in tables using Conditional Random Fields with 99.55% classification accuracy• Implemented ranking and retrieval of ontology classes for keyword search queries using different augmentation techniques including word2vec, WordNet and web search.• Proposed and implemented a canonical representation of physical unit strings using QUDT ontology.

• Lead developer in building a web analytics application used by 1000+ publishers. Responsible for end to end design, development, and deployment of the system.• Built a distributed system using Cassandra to handle 10k+ QPS incoming data streams.• Designed an ETL pipeline to process data from Cassandra, and HDFS for use in Spark queries.• Built a distributed service-oriented architecture to fetch live web pages, take screenshots, and construct heat maps from visitor activity data stored in Cassandra.• Implemented a JavaScript module to collect visitor metadata and behavioral data with support across multiple device types (mobile/tablet/desktop).• Created a prototype to consume tweets in real time and identify user sentiments for different brands.