Aditya Singh — Enterprise AI portfolio

This repository publishes a static, evidence-led portfolio for enterprise AI, banking transformation, governed architecture and delivery leadership. GitHub Pages serves the files directly from main; no framework build or runtime service is required.

Information architecture

• index.html is the recruiter- and executive-oriented portfolio entry point.
• build-evidence.html routes to presentation, architecture and code evidence.
• case-study-library.html preserves all 30 evidence-qualified case-study decks.
• technical-projects.html combines curated flagships with the complete technical-project index.
• technical-concepts.html preserves source-incomplete concepts with explicit evidence gaps rather than presenting them as implemented systems.
• samples/technical-projects/<slug>/ holds each public technical brief, architecture pack and sanitized source download.
• samples/presentations/ and P Presentations/ contain customer-neutral samples and browser-friendly case-study evidence.
• presentation-archive.html keeps the complete authorized 16-set archive discoverable through release-backed downloads and a SHA-256 manifest.
• assets/ contains shared public artwork, including the 1200 × 630 social card.

The social card keeps an editable SVG source and a committed PNG used by link previews. On macOS, regenerate the raster after an SVG edit with:

sips -s format png assets/social-preview.svg --out assets/social-preview.png

Preservation rule

Published content is preserved. Curation is implemented through navigation, featured views, collection pages, filters, accordions and archive routes—not by deleting projects, slides, diagrams, source packs or supporting evidence. Every artifact should retain a clear route from a complete-library page.

Private source inputs are not published. Only reviewed, sanitized outputs belong in the public technical library. Credentials, personal data, customer datasets, deployment state and local filesystem paths must never be committed.

Regenerating technical evidence

The technical portfolio generator treats the locally held, authorized project archives as untrusted input. It selects allow-listed text files, redacts sensitive material, creates architecture SVGs and emits the public project pages and ZIPs. It never executes project code.

python3 scripts/build_technical_portfolio.py

Review the generated diff and the source-pack checksums before publication.

Sitemap and local quality checks

The sitemap is derived from canonical URLs, so local working HTML without a public canonical is never exposed automatically.

python3 scripts/generate_sitemap.py
python3 scripts/generate_sitemap.py --check
python3 scripts/site_qa.py

The dependency-free QA checks internal links and fragments, page landmarks, skip links, H1 count, duplicate IDs, image alternatives, canonical/Open Graph/ Twitter/JSON-LD metadata, sitemap coverage, the social-card dimensions and GitHub file-size limits. Assets above performance thresholds are reported as warnings while GitHub-incompatible files fail the check.

The same checks run in .github/workflows/site-qa.yml for pull requests and pushes to main.

Publishing

GitHub Pages should be configured to deploy the repository root from main. The .nojekyll marker keeps the static asset tree intact. After deployment, verify the homepage, one technical detail page, one archive download, the sitemap and an intentionally missing URL that exercises 404.html. The release-backed archive is part of the public information architecture: moving large binaries out of the Pages tree must never remove their visible routes, download links or checksum evidence.