AI project roadmap

Practical tools for the work I already understand.

These projects use synthetic data and fictional organizations. They are designed to demonstrate secure implementation, measurable usefulness, and human accountability—not unsupported AI claims.

Project 01 · In development

Vulnerability Report Assistant

A security workflow that turns synthetic scan findings into risk-ranked remediation plans and management-ready summaries.

What it demonstrates

Severity plus asset-context prioritization
Separation of evidence from AI recommendations
Human approval for remediation decisions
Security and AI governance considerations

Planned stack

Python, synthetic vulnerability exports, CVSS-style scoring, reporting interface, documented controls.

Project 02 · Planned

AI Help-Desk Triage

A human-in-the-loop assistant for classifying synthetic requests, identifying urgency, and preparing concise technician summaries.

What it demonstrates

Ticket classification and priority recommendations
Technician review before any external action
Evaluation against a labeled synthetic dataset
Clear audit history for every recommendation

Planned stack

Power Automate or Python, structured synthetic tickets, evaluation dashboard, approval workflow.

Project 03 · Planned

Manufacturing IT Knowledge Assistant

A retrieval-based assistant that answers support questions from a controlled set of fictional SOPs and troubleshooting guides.

What it demonstrates

Cited answers grounded in approved documents
Role-aware access and data boundaries
Refusal when the source material lacks an answer
Evaluation for accuracy, citation quality, and unsafe responses

Planned stack

Python, document retrieval, a language-model API, lightweight web interface, synthetic manufacturing IT documentation.