Toffee
Project Toffee aims to enable automated program repair by leveraging neuro-symbolic programming, which combines program analysis methods with the latest advancements in large language models (LLMs).
Toffee
Toffee
Project Toffee aims to enable automated program repair by leveraging neuro-symbolic programming, which combines program analysis methods with the latest advancements in large language models (LLMs).
Project Overview
The overall goal of the project is to reduce the manual effort required for bug localization, bug reproduction, and program repair. The objective is to develop human-in-the-loop solutions that minimize the manual tasks involved in typical bug-fixing processes as much as possible. Project Toffee aims to leverage neuro-symbolic programming, combining program analysis with large language models (LLMs) to automatically fix bugs—starting with simple bug fixes and progressing to more complex bugs that require thorough program analysis.
Mahin Chandramohan is leading AI/ML initiatives in the Intelligent Application Security Group at Oracle Labs, Australia.
His research interests include:
• Automated program repair (APR)
• Large language models (LLMs)
• Open Source Intelligence (OSINT)
• Threat intelligence
• Malware analysis and binary reverse engineering
• Machine learning for program analysis and bug detection (MLonCode)
Education:
• PhD in Computer Science, Nanyang Technological University, Singapore
• B.Eng. in Computer Engineering, Nanyang Technological University, Singapore