Jon Bisila

Principal Investigator

Building an AI-powered platform that uses frontier models to parse threat intelligence reports and generate structured, actionable outputs for security analysts. Collaborating with UI/UX specialist on workflow integration.

Lead Developer

Developing an evaluation framework for LLM-assisted reverse engineering, using Claude Code with custom skill definitions and PyGhidra MCP server integration. Exploring how well current models can support binary analysis workflows.

Principal Investigator

Led a team developing a security-focused benchmark for evaluating LLM capabilities in C-to-Rust translation. Designed evaluation methodology targeting memory safety vulnerabilities (CWE-based test cases) to assess whether models can eliminate security weaknesses while preserving functional correctness. Transitioned project leadership for continuation.

Primary Contributor

Contributed to red team assessments of production ML systems, evaluating vulnerabilities across adversarial ML and model performance attack surfaces. Focused on defining evaluation criteria and documenting failure modes that could be exploited.

Lead Research Engineer

Built production pipeline using specialized LLMs for domain-agnostic entity extraction and translation across multiple languages, automating previously manual analysis at web-scale. Led experimental validation demonstrating custom LLM approach outperformed off-the-shelf solutions for requirements.