Ashfaaq Farzaan

Blog Posts

CyberSOCEval: A Surprising Lesson from the Malware Reasoning Benchmark

Exploring how input structure affects LLM reasoning in cybersecurity, using JSON, TOON, Graph, and Markdown representations.

Read Article

Inside ORACLE: How AVA Triages Alerts Like a Senior Analyst

An inside look at AVA's hypothesis-driven triage framework: how it reasons like a senior analyst across six stages, and gets sharper with every alert it sees.

Read Article

Cyber LLM Benchmark Hub: A Home for Measuring AI in Cybersecurity

A central place to compare cybersecurity LLM performance across tasks, domains, and models. The hub brings together 35 benchmarks, 78 models, and 10 categories.

Read Post

The Third Verdict

Exploring the nuances of AI agent decision-making and accountability in autonomous cyber operations.

Read Article