AI Safety — Bioinformatics

Efe Akça

I am a researcher in AI safety and cancer metabolomics. My interests: pragmatic interpretability, metabolic vulnerabilities in cancer, abstract mathematics.

GitHub ↗ Scholar ↗ Publications CV

01 Research

Measuring Probe Performance in Cybersecurity

Is code vulnerability represented in model activations? What do probes trained on vulnerable code detect?

Jun 12, 2026

Nothing here yet - my research writings will be here.

All research →

02 Notes

Nothing here yet - short-form writing will be here.

All notes →

03 Projects

probes-nlas-cyber Linear probes and natural-language activations for cyber-vulnerability classification on Gemma 3 jellip A delightful Japanese practice app

04 Links

GitHub. Google Scholar and publications. CV.