> cat ./blog/index.md

Articles & Insights

Thoughts on AI/ML security, agentic AI, and the evolving threat landscape. Published on LinkedIn.

From Monolith to Micro-Agents: My 4-Stage Journey in Building Secure Local AI Systems

An evolutionary progression through four architectural phases for building a local AI system, from a single monolithic agent to a distributed multi-agent network with security safeguards at each stage.

AppSec Using Agentic AI: Running SCA Scans via Claude CLI Skills

Using Claude Skills to automate security scanning workflows, specifically implementing Software Composition Analysis (SCA) scans through the Bandit tool via Claude CLI.

Why "AI Skills Developer" Will Be the Breakout Role of 2026

An emerging specialized role combining software engineering with LLM orchestration, where prompt engineering evolves into a structured discipline using Claude Skills, Sub-agents, Hooks, and MCP.

Secure Source Code by Design with Cline: A Practical Guide for Engineering Teams

A methodology for configuring Cline (VS Code AI extension) to enforce OWASP security standards during code generation, combining Memory Bank, .clinerules, and modular secure coding standards.

AI/ML Security: Old Attacks, New Targets

Threats targeting AI systems aren't fundamentally novel — adversaries use the same tricks (injection, tampering, leaking secrets) but now focus on training datasets, models, and AI outputs.

Evolving Protocols Around Agentic AI: MCP, A2A, ACP, AGNTCY

Examining emerging standards for agent-based AI systems including MCP, A2A, ACP, and AGNTCY protocols anticipating an Internet of Agents (IoA).

The 3 Phases of AI Mastery: Skills, Tools, Roles, and Privacy Practices

Three distinct AI proficiency levels — from end-user to ML specialist — along with corresponding competencies, recommended tools, and security considerations for each stage.

ML Cyberattacks Taxonomy

A comprehensive framework for understanding cyberattacks targeting machine learning systems, organized across training data, model, and inference stages.