Security Researcher: AI Vulnerability & Benchmarking

Hacktron; USA; Hybrid; USD 120.000 - 180.000; Vollzeit

Benefits dieser Stelle

Hybrid work

Coworking memberships

European offsites

Equity

Health, Vision, and Dental Insurance

Zusammenfassung

Hacktron in the United States is seeking a researcher to study and test security tools that evaluate how AI agents identify vulnerabilities in real software. You will merge hands-on vulnerability research with benchmark design and product testing in a small, high-ownership team.

You’ll collaborate with research and engineering colleagues across international locations, expanding vulnerability research on open-source software and creating benchmarks to measure agent performance on security tasks.

Qualifikationen

• Demonstrated strength in at least two relevant areas.
• Understanding of reachability analysis or taint tracking.
• Knowledge of web application vulnerabilities demonstrated through security competitions or discoveries.
• Experience presenting technical security research at conferences.
• Experience using both conventional security tools and language-model-based tools in research.

Aufgaben

• Expand vulnerability research on open-source software, including work alongside a founder.
• Create realistic benchmarks to measure agent performance on security tasks.
• Observe agents operating on real systems and use findings to improve their effectiveness.
• Assess products and internal systems for security weaknesses.
• Share research findings with engineering to inform product development.

Kenntnisse

Reachability analysis

Taint tracking

Web application vulnerabilities

Security research presentations

LM-based security tools

Conventional security tools

Jobbeschreibung

Hacktron in the United States is seeking a researcher to study and test security tools that evaluate how AI agents identify vulnerabilities in real software. You will merge hands-on vulnerability research with benchmark design and product testing in a small, high-ownership team.

You’ll collaborate with research and engineering colleagues across international locations, expanding vulnerability research on open-source software and creating benchmarks to measure agent performance on security tasks.