This project aims to construct better log embeddings to improve the performance of log-related downstream tasks (e.g. anomaly detection, log analysis).
Advised by Professor Mike Cafarella
This project aims to extend software/ program analysis tools from a single function to a state machine.
Indepdent project; Advised by Professor Oleg Sokolsky
This project aims to tackle the prolonged reinferencing time upon retraining of ML models. One of the applications is the incremental maintenance of entity matching in knowledge graphs and associated query results.
Independent project; Advised by Professor Zack Ives
Devised the algorithm that uses existing inference results as threshold for subsequent inference results
Adapted techniques from relational databases such as incremental view maintenance to graph databases
JUpyter Notebook with Enhanced Access and Understanding (JUNEAU)
JUNEAU is a system of holistic data management tools which can find, standardize, and benefit from the existing resources in the data lake. With much improved scalability, the latest version of Juneau supports searching hierarchical and joined data as well as detecting composite data profiles.
Advised by Professor Zack Ives
Developed the recursive parsing algorithm that normalizes hierarchical data into base relational tables
Initiated the use of sketching techniques to speed up the process of creating and matching data profiles
Enhanced scalability through pre-computations, incremental techniques, and in-database implementations