• Built and evaluated search ranking models, two of which were in the top three in terms of CTR improvements in a year.• Transformed vision for a query clustering project into a neat prototype that was used by many cross-functional teams like search relevance, growth, and news feed integrity.• Improved the quality of the knowledge graph using my knowledge of Wikipedia data.• Conducted lots of interviews, mentored new hires, and managed interns.

• Implemented features and built probabilistic models for personalized recommendations.• Integrated neural network models with the ranker plugin of Elasticsearch.• Designed a system using Elasticsearch for generating concept summaries for a document, which cut the development time by 2 months.

• Created an ML-based automatic test generator to generate a test from the job description and problem database.• Designed and implemented a ranker for ranking resumes for a given job description.• Presented JyMob’s vision to accelerators and VCs.

Conceptualized query graph and outlined a vision for its uses in spelling correction, query tagging, and search suggestions.

• Migrated site’s proprietary search index to Solr, increasing quality and reliability and reducing the number of servers required by 50%.• Designed and implemented an ML-based recommendation system and spelling correction system for music queries in various Indian languages.• Guided search editors for testing search results.

• I led the ML side of the team in building Glue Pages, which was praised by experts (http://anand.typepad.com/datawocky/2008/05/why-yahoo-glue.html). • I designed and implemented query classifiers for various dimensions such as topic, user intent, and geo-location.• Also built models for selecting and ordering Glue modules on topic pages.• Implemented glue data processing on Hadoop in order to speed it up.• I worked on various tasks such as disambiguation of queries, editorial testing, etc.

Worked on various Search and ML projects such as video recommendations, channel theme detection, filtering, finding user query intent, and reconstructing queries from partial queries as part of the media search team.

• Worked on various search query log analysis projects. Optimized generation of concept networks based on search query logs.• Implemented prototypes for various concept network applications such as query suggestion, clustering, data generation for spelling correction, query categorization, and seasonality detection.