Skip to main content

AI Use Disclosure

Last updated: 27 May 2026

This page documents how artificial intelligence tools are used in building, maintaining, and operating this portfolio site. I believe in transparency about AI involvement.

1. Site Construction

This portfolio was built with AI coding assistance (Google Deepmind's Antigravity / Gemini models and Anthropic's Claude). AI was used to generate code scaffolding, debug issues, write boilerplate, and suggest architectural patterns. All final decisions, content, project descriptions, and design choices were made and reviewed by me.

2. Project Content Assistance

The admin panel includes an AI-assisted feature that can generate project descriptions, Challenge/Action/Result summaries, and skill tags from uploaded project reports. Specifically:

  • PDF reports are sent to Google Gemini for analysis and summarisation.
  • AI-generated text is reviewed and edited before being published.
  • All factual claims about project outcomes are verified against source documents.
  • AI-generated C/A/R sections are marked for review ("estimated") when applicable.

3. SEO & Taxonomy

AI tools (Gemini, Groq) are used to generate SEO meta titles, meta descriptions, and skill tag suggestions. These are always reviewed before publication and treated as starting points, not final copy.

4. Blog Content

Blog posts and "Thoughts" on this site are written by me. AI may be used for editing, grammar checking, or brainstorming, but the ideas, arguments, and conclusions are my own. Posts are not AI-generated.

5. Automated Systems

This site includes a Career Discovery engine that autonomously discovers and evaluates job listings. This system:

  • Uses LLMs to score job fit against a candidate profile (for private admin use only).
  • Does not make any autonomous hiring applications; it only surfaces and scores listings for human review.
  • Is not visible to site visitors.

6. AI Providers Used

  • Google Gemini: report analysis, C/A/R generation, SEO optimisation.
  • Groq / LLaMA: fast inference fallback for taxonomy and tagging.
  • Anthropic Claude: code assistance during development.

Data sent to AI providers is limited to: project report text, project metadata, and short prompts. No visitor personal data, subscriber emails, or analytics data is sent to any AI provider.

7. Machine-Readable Profile

A machine-readable version of this portfolio is available at /llms.txt. It is designed to provide accurate, up-to-date context to AI crawlers and LLM-based tools.