Using AI to build valuable products for LinkedIn's Premium members

• Led a revamp of LinkedIn Premium's Job Recommendation module called Top Applicant Jobs using Multitask Learning and Graph Neural Networks which has led to tens of millions of dollars in additional revenue so far

• Leading a revamp of the Notifications experience for Premium members to deliver value

• Leading the data science efforts for Expedia’s first revenue management tool for hoteliers

• Built hotel demand prediction models with less than 5% error using R, Python, Hadoop (Hive)

• Reviewed development of the production version of the hotel demand prediction model

• Regularly presented modeling results to senior management and product managers

• Led the first-ever user-id based hotel search personalization initiative at Orbitz

• Built recommender models for hotel search using Python, Spark (MLlib), Hadoop (MapReduce, Hive)

• Performed A/B test and increased NDCG (ranking quality) of hotel search by 10%

Fundamental to the field of quantum chemistry is the Hartree-Fock minimization problem which has numerous local minima, due its inherent non-convexity. In that context, the following contributions were made.

• Published 3 first-author papers in top journals – Physical Review A and The Journal of Chemical Physics

• Formulated a convex relaxation to the Hartree-Fock minimization problem in quantum chemistry

• Developed the first-ever algorithm to obtain upper and lower bounds to global Hartree-Fock minima

• Implemented libraries in C++ and Python for scientific computations on Linux machines

• Published a research paper in the Journal of Molecular Structure

• Implemented numerical methods in C++ for computational modeling of physical/chemical systems