Apr 2015 – Jun 2017 · 2 yrs 2 mos · San Francisco Bay Area

• Technically led various efforts on improving search results relevance for the Apple Maps Search local engine. The following projects/achievements were noteworthy:
• 1. Significant focus on User/Query logs to extract potent features helping in Query Intent/Understanding
• 2. Several Techniques to compute click through rate features (CTRs) to inspect user queries from multiple view points
• 3. AB Tested Several Machine Learned Models to Improve Query Intent Prediction with clear winners getting 100% prod deployments in successive releases
• 4. Exploiting Query Transition in User Sessions to extract Synonyms, Query Variants to improve Recall
• Overall, worked on several aspects of improving relevance of a Local Search Engine. The Apple Scale, Queries from 100s of countries in various languages, Frequent Data changes, multiple teams as stakeholders in some projects .... all these made the experience very exciting!

Mar 2011 – Mar 2015 · 4 yrs

• > Devised Algorithms to grapple with 100s of gigabytes of Netflix users' search/play data to extract relevant bits & use them to improve search experience of millions of Netflix Customers.
• > Developed & Improved relevance of the Search Engine at Netflix to enable Netflix members find the movie they want & give helpful recommendations to enable discovery.
• > Like most of Netflix, the Search engine was also a cloud based system with high availability and low latency requirements. Helped create & improve monitoring infrastructure to granularly & continuously keep an eye on various Search services used in production environment.
• > Created various features from users' search history and provided them as a service to other teams. -> Helped in internationalizing the search as the Netflix service was launched in various countries (Latin America, Great Britain, France, Germany, Nordics)

Nov 2009 – Mar 2011 · 1 yr 4 mos

• a) Search Engine Development
• YP.com is a local search engine, rewritten from scratch during my stay there. I was involved in its design/development/shipping since inception. We used SOLR as the base search engine & made it highly configurable to account for frequent modifications of business logic by product & sales teams. We did significant work on tuning the system both relevance & performance wise (using hierarchical indexes, Solr multicores, query-frequency analysis, Geo-sharding, multiple caching layers, ...).
• b) Query Categorization System
• I co-developed a query categorization system which would assign a user query one or more concepts from a taxonomy. It was based on YP past click logs, category taxonomy, category keywords and document names. The Java based system consisted of many modules arranged in a hierarchical manner and was heavily multi- threaded.

Feb 2008 – Oct 2009 · 1 yr 8 mos

• Developing Web 2.0 and Social Networking centric Software for Recommendation Systems & Digital Libraries

Jun 2007 – Sep 2007 · 3 mos

• I worked on developing a Cross Lingual Sentence Alignment Tool during my internship. The tool was developed successfully and was accepted by the XRCE scientists.

May 2007 – Sep 2007 · 4 mos

• I worked as a Google Summer Code intern for three months working on a Context Free Grammar Learning Project. The project was successfully completed in time and was accepted by Google. I also extended this project in my Master Thesis work at Johns Hopkins University.

Jun 2004 – Jul 2006 · 2 yrs 1 mo

• VOIP Solutions for Telecom products including SIP Application server, Prepaid applications