Building a new product that helps small businesses automate their needs with scalable backend infrastructure and ML solutions. Feel free to ping me if you are interested in being part of the adventure.

Vignanam apps and the website, serving 1M+ monthly active users, showcase the best of Indian Cultural Knowledge across multiple scriptures, authors, and festivals in 20+ Indian language scripts. The apps serve as a spiritual daily companion, offering rich details on upcoming Indian festivals with a detailed Indian calendar and the hymns to chant at each festival.

Led the Ads Ranking efforts at Twitter. Managed 50+ ML engineers and SWEs with multiple managers and Senior EMs leading various product modeling and infrastructure teams. Our mission is to surface the best experience to the users and advertisers through the most relevant ads. We pushed the boundary for machine learning in terms of advanced ML models for predicting user engagement with one of the large-scale production systems in the industry for deep learning and online learning.

NLP and Machine Learning PlatformEarlier: Personalization & User Modeling Sciences

Web Document Understanding RelevanceManaged a team of scientists in the core ranking team at Bing, focusing on Document Understanding Relevance and ambitious Question Answering (like "what is the average typing speed?" and "what car did James Bond drive in golden eye?"). - Led the efforts in natural language understanding of text including named entity recognition, document summarization and fact extraction.- Don't have time to read a full web page? Try Skype's new Summarize bot that we built at https://bots.botframework.com/bot?id=4104c73d-90d6-4572-95eb-95241bd0cacb.Bing PredictsManaged the team and built Social Pulse platform for Bing Prediction engine (http://www.bing.com/explore/predicts), which enabled key predictions including sports predictions like NFL predictions for 2014-15 season, NBA draft predictions, political elections like Scotland referendum predictions and reality TV show elimination predictions for American Idol and The Voice. Semantic RankingImproved web ranking for tail and semantic queries at Bing using key features like domain popularity and ngram language models and topic models.Web Search Core Ranking Team, BingBuilt a platform to build and refresh fresh ranking features from the social data, including Twitter and Facebook data to help improve ranking for fresh queries at Bing.Reference Answers and Question Answering @ BingBuilt the factoid triple extraction platform to power the best question answering sytem for factual questions like "how old is einstein?" and "how much does mark zuckerberg make?"

Technical Lead - Object Ranking & RecommendationLed all Object Ranking related products at Yahoo! search and structured data based search experiences. Built recommendation systems for related videos and related Finance ticker symbols. Co-hosted a tutorial for object ranking at CIKM 2011.Technical Lead - News ClusteringLed the Yahoo! News Search Results Clustering project across multiple teams to cluster the search results for news queries and helped deliver the experience to Yahoo! news search. Won You Rock! award for the this effort.Research Scientist - Yahoo! Web Search Ranking Team- Machine Learned Ranking Modeling & Multi-task learning- Information Extraction and Page structure analysis for web search- Region classification for web pages - Regional sensitivity identification of queries to improve ranking for local intent queries.

Analysis of the quality of the search index and devising metrics to identify and investigate discrepancies in the entire Web index data.

Query rewriting with automatic suggestions based on synonyms and co-occurrence of query terms to improve the relevance of the search results.

Built a predictive modeling system that analyzes health insurance member flow and and forecast claim costs and health conditions with a Bayesian Vector Autoregression model.