Helped design and launch lead generation system to grow real-estate business utilizing react and python. Experimented with recommendation system to better target leads.

Trained a page visitation model based on the skip-gram architecture to incorporate co-visitation, location, and category signals in page recommendation and launched generated embeddings as candidate source. (C++, Apache Hive)Trained a topic signal based page recommendation model using sparse neural network architecture to recommend users with similar pages based on their ad-interest and achieved conversion rate two times higher than the average in the online experiment. (Caffe 2, Pytorch) Created embeddings used in recommendation systems for candidate generation, features in ranking layer optimization, and parameter initialization in transfer learning.

Computational Simulation and Statistical Analyze of Protein Folding in Confined Media. (C, OpenMPI)Wrote program in OpenMPI to study the stability and configuration property of protein in various confined media using discontinuous molecular dynamics(DMD) simulation.Effect of the geometry of confining media on the stability and folding rate of αα-helix proteinshttps://doi.org/10.1063/1.5020841