Open to opportunities

Jason Phillips

@jasonphillips

Message

I build adversarial AI evaluation and offensive security systems with auditable evidence chains.

United States

Message

What I'm looking for

I'm looking to build and deploy adversarial AI security, LLM evaluation, and offensive security systems where findings are repeatable, auditable, and grounded in measurable evidence.

I've built LLM evaluation workflows at Labelbox and led offensive security work at Apple, spanning scientific model grading, adversarial red teaming, API security reviews, and repeatable evidence-driven security testing.

At Apple, I identified authentication and authorization bypasses, chained misconfigurations into end-to-end compromise paths, discovered a Kubernetes build-pipeline container escape, and reduced AWS IAM privilege sprawl. I also reverse engineered macOS and Apple Silicon internals, analyzed macOS malware, and built static-analysis and Python automation for internal red team workflows.

I build security and AI systems that make results verifiable: from Prism's tamper-evident agent execution ledger to Sentinel's autonomous, fail-closed security-testing platform. My background in biomedical engineering, applied mathematics, scientific computing, and materials informatics helps me evaluate technical model output across biology, chemistry, physics, and medicine.

Experience

Work history, roles, and key accomplishments

Senior AI Evaluation Engineer

Labelbox

Apr 2024 - Feb 2026 (1 year 10 months)

Designed and executed LLM evaluation workflows across coding, mathematics, physics, chemistry, biology, medical, and multimodal model categories, applying graduate-level domain knowledge to grade scientific reasoning. Authored subject-matter rubrics for quantitative science tasks and built tooling to normalize model outputs and generate audit-ready reports.

LLM Evaluation Python TypeScript Adversarial Testing Data Normalization Audit Reporting

Education

Degrees, certifications, and relevant coursework

University of California, Davis

Master of Science, Biomedical Engineering

Graduate coursework in physiology, biomechanics, biomedical modeling, signal processing, control systems, numerical methods, and scientific computing, with emphasis on quantitative validation.