Sanjay Nagraj

Title: Sanjay Nagraj and LLM Security Innovation

Introduction

Sanjay Nagraj is an inventor based in Dublin, Ireland, and works at Harness Inc. His patent work addresses security risks associated with large language models (LLMs), including harmful outputs, vulnerable plugins, and exposure of private data.

Latest Patents

Nagraj is associated with one patent: System and Method for Preventing Exploitation of Large Language Models. The disclosed method analyzes API calls, including user requests and LLM-generated responses, to identify potential vulnerabilities. It uses contextual analysis to flag suspicious calls, assigns toxicity scores to responses, assesses plugins for malware risks, and applies trained models to detect private-data leaks. A vulnerability detection model combines these findings into a risk assessment, which can guide mitigation actions such as blocking flagged calls or harmful responses, restricting access to sensitive information, and alerting users to potential leaks.

Career Highlights

Nagraj’s patent reflects work on practical safeguards for systems that use large language models. The invention brings several security checks together to support risk assessment and mitigation.

Collaborations

The provided information identifies Jisheng Wang and Naresh Agarwal as Nagraj’s coworkers. Their specific contributions to the patent were not provided.

Conclusion

Sanjay Nagraj’s patent focuses on a layered approach to detecting and mitigating security risks in large language model systems, including suspicious API activity, harmful content, plugin vulnerabilities, and private-data leaks.