Partnered with Bonfire Interactive Ltd. on a consultant project.

● Designed an algorithm to detect sensitive key words to an entity according to their

similarity scores using the vector embedding of the domain vocabulary.

● Built an application to extract identifying information of entities from records and

anonymize them, enabling Bonfire to unlock a feature of automatic configuration

templates for procurement projects.

● Hands on data aggregation and munging through Postgresql/Pandas and use natural

language tools like LDA, TF­IDF for topic/keywords extractions.

● Built a recommendation system for Bonfire to automatically extract essential criteria of

procurement projects based on historical records.

● Designed an algorithm to parallelize serial simulation of lattice gravity using OpenMP/pthread in C++ and achieved more than 6x speedup on 8-core clusters and machines with Intel Xeon Phi coprocessors.

● Implemented the parallel tempering algorithm in Monte Carlo simulation of lattice

gravity using MPI to improve the sampling at critical parameters.

● Participated as a Co-PI in the kaon-decay lattice simulation using the Blue Waters HPC clusters at NCSA; hands on managing and transferring large datasets among HPC systems and performed large scale simulations.

● Directed data analysis of a B decay simulation project; improved precision by 50%.

● Wrote Python modules for Bayesian fits; invented a functional curve fitting method.