Professor Song received the NSF CAREER award for his work on "scalable concolic execution." Software vulnerabilities are major threats to cybersecurity. Concolic execution (a.k.a., dynamic symbolic execution) is a powerful program analysis technique to find vulnerabilities. However, its potential has not been realized in the context of testing real-world software due to the poor scalability of traditional concolic executors. This project aims to advance the scalability of concolic execution via machine-learning-guided path pruning and fast path constraint collection and solving.