SPDZCoder combines expert knowledge with Large Language Models to automate the generation of privacy computing code, addressing a major challenge for developers.

• Creates a specialized system that translates standard Python code into privacy-preserving formats
• Leverages both programming patterns from experts and LLM capabilities for optimal results
• Outperforms general LLMs in generating correct, efficient privacy computing code
• Lowers the technical barrier for implementing privacy-preserving computation

Security Impact: Enables wider adoption of privacy computing frameworks by automating complex code translation, helping organizations implement stronger privacy protections without specialized expertise in cryptographic protocols.

SPDZCoder: Combining Expert Knowledge with LLMs for Generating Privacy-Computing Code