Collaborated with Microsoft Research to build ThreatKG, an AI-powered platform that constructs a threat knowledge graph for automated cyber threat knowledge gathering and management using open-source cyber threat intelligence (OSCTI) data
•
Designed and implemented a robust multi-threaded OSCTI report crawler framework (Python) to crawl from 40+ major security websites collecting a total of 143K+ reports and operating at a crawling rate of 325+ reports/second•Trained and deployed machine learning models (SKLearn) to preserve threat related articles predicting with 90+% accuracy and F1
•
Built a Bidirectional LSTM-CRF neural network using Pytorch to recognize security entities (e.g. malware, bad actors) achieving 95% accuracy and F1
•
Lead Developer for ThreatKG UI web app (React, D3.js, Flask) to facilitate threat analysis and exploration from Neo4j graph database containing 347K+entities and 1.7M+ relations; includes diverse searching capabilities through Cypher queries and Elasticsearch
•
Contributed to research paper writing leading to publications at SIGMOD 2021 and VLDB 2022
