Arcanna

    Authors

    Alina Marcu, PhD

    Chief Data Scientist at Arcanna

    Alina is the Chief Data Scientist at Arcanna and the original author of the Decision Layer, the per-tenant, deterministic model that grounds every AI verdict in the team's own trained judgment.

    She built the governance mechanisms that make Arcanna's decisions auditable: confidence thresholds, drift detection, rollback, and full traceability of every analyst's feedback. Her position is that in mission-critical operations, governance isn't a layer you add later: it's the architecture.

    "LLMs are great assistants because they have broad knowledge across many topics. But in mission-critical operations, you cannot have a system making decisions that are unverifiable, unaccountable, or hallucinated."
    Alina Marcu, PhD
    Featured23 min watch · 8 min read
    SME Interview

    Why AI in the SOC Needs Governance

    AI in the SOC isn't governed by default. That's the real risk. In this conversation, Alina Marcu — Chief Data Scientist at Arcanna and the original author of the Decision Layer — makes the case for why mission-critical security operations need a Trust Layer between telemetry and automation. Deterministic. Auditable. Trained on your team's institutional knowledge, not internet opinions.

    Full transcriptVideo + written
    Watch the Interview →

    Recent Articles

    AI in a Google SecOps SOC: Faster Triage Without Rewriting Your RulesGoogle SecOpsJul 9, 2026 · 8 min read

    AI in a Google SecOps SOC: Faster Triage Without Rewriting Your Rules

    Adding AI to Google SecOps cuts MTTR only if the decision underneath is reproducible. How to make fast triage safe — no detection rules rewritten.

    Alina Marcu, PhD
    Alina Marcu, PhD · Chief Data Scientist
    Predictable Alert Triage at Scale: Why SOC Consistency Beats Raw AutomationAlert TriageJun 15, 2026 · 8 min read

    Predictable Alert Triage at Scale: Why SOC Consistency Beats Raw Automation

    Adding automation to inconsistent triage just scales the inconsistency. Why predictable alert triage — same input, same output — is the precondition for safe SOC automation.

    Alina Marcu, PhD
    Alina Marcu, PhD · Chief Data Scientist
    What to Look for in AI SOC Platforms: Thresholds, Drift, RollbackDecision LayerMay 20, 2026 · 7 min read

    What to Look for in AI SOC Platforms: Thresholds, Drift, Rollback

    How do you evaluate an AI SOC platform that will still be working in 18 months? A practitioner-led framework: thresholds, drift detection, and rollback.

    Alina Marcu, PhD
    Alina Marcu, PhD · Chief Data Scientist
    Governing AI in the SOC: The Case for a Trust LayerAI GovernanceApr 28, 2026 · 9 min read

    Governing AI in the SOC: The Case for a Trust Layer

    How to make AI decisions auditable, deterministic, and defensible to the CISO.

    Alina Marcu, PhD
    Alina Marcu, PhD · Chief Data Scientist

    See Arcanna in the SOC

    Decision models and agentic workflows, built on your team's own judgment.

    Get a Demo →