Independent AI Risk & Governance Consultant

I conduct evidence-based risk and readiness reviews for GenAI, RAG, and agentic systems.

My work connects:

Architecture → Risk → Consequences → Controllability

I examine what is supported by available evidence, what remains unverified, and what should be addressed before deployment, scaling, client handoff, or increased autonomy.

A pre-deployment or design-stage review of system architecture, workflow boundaries, authority, permissions, retrieval, human approval, failure handling, and expected operating conditions.

A runtime-oriented review using available operational evidence such as demonstrations, configurations, sanitized logs, traces, test results, and observed system behaviour.

Either route may be evidence-enriched. Technical evidence supports professional judgment; it does not automatically produce findings, scores, or readiness decisions.

• RAG source and metadata boundaries
• retrieval behaviour and evidence traceability
• agent authority and tool-use controls
• workflow telemetry and reconstruction
• failure handling, recovery, and controllability
• evidence quality and limitations

The AI Systems Risk & Evidence Lab develops controlled experiments and bounded technical evidence for RAG, agent, observability, and operational AI risk questions.

Lab outputs are evidence inputs. They are not formal audits, certifications, legal conclusions, penetration tests, or substitutes for production evidence.

• Website: https://borisabuzov.com
• Services: https://borisabuzov.com/services/
• Evidence: https://borisabuzov.com/evidence/
• LinkedIn: https://www.linkedin.com/in/boris-abuzov-854176426
• Contact: https://borisabuzov.com/discuss/