Skip to content
Sujan Ghimire
Research
Ongoing research

Agentic LLM-Driven Hardware Logic Obfuscation

Exploring LLM-assisted and agentic approaches to hardware logic obfuscation and intellectual-property protection.

  • Research
  • AI/ML

Problem

Logic obfuscation can protect hardware intellectual property from piracy and reverse engineering, but applying it correctly requires specialized expertise. It is an open question whether LLMs and AI agents can perform this task reliably and securely.

Technical approach

  1. 01Investigate frameworks in which LLMs apply logic obfuscation to hardware designs.
  2. 02Evaluate agentic LLM workflows for IP protection.
  3. 03Assess both the functional correctness and the security of the obfuscated designs.

My contributions

  • Lead author of the IEEE VTS 2026 paper evaluating agentic LLM-driven obfuscation for IP protection.
  • Co-author of the IEEE DCAS 2024 paper on an automated hardware logic obfuscation framework using GPT.

Associated publications