A lightweight, static portfolio connecting AI evaluation, data operations and practical software delivery with live demonstrations and documented case studies.
- Professional introduction and experience summary grounded in Dhairya's supplied work history.
- A three-service Delivery Lab: chatbot QA, responsive landing pages and technical SEO audits, each linked to the relevant existing Fiverr offer.
- Interactive authored QA comparisons, genuine desktop/mobile screenshots and filterable fictional SEO findings.
- A downloadable ShopDesk v2 QA sample: six-page PDF report and editable Excel scorecard, with ten fictional cases, a separate scoring contract and proposed acceptance checks. No live model or client work is represented.
- A local-only brief builder with preview, copy, text download and stale-draft invalidation. Brief entries are not submitted or stored by the page.
- Live links to Fieldwork, its Planning Lab, Fieldnote Studio and the Python LLM Evaluation Lab.
- Actual project screenshots, not fabricated client work.
- Contact through existing public LinkedIn and Fiverr profiles.
- Responsive navigation, keyboard handling, reduced-motion support and descriptive image text.
- Visible Light/Dark controls, dark by default. Only the selected appearance is saved in this browser; blocked storage does not prevent switching.
The portfolio projects are independent, AI-assisted demonstrations. Their examples are fictional; the Planning Lab is rules-based, not a live AI agent. No employer source code or client records are included. The evaluation examples and scores are authored synthetic material, not measured model performance. The issue sheet is illustrative: no live website crawling or SEO measurement occurs. Prices and delivery commitments are not invented; buyers are directed to Fiverr to confirm current packages and agree scope. Fiverr-originated conversations and payments should remain on Fiverr.
This site uses plain HTML, CSS and a small browser JavaScript module. There are no production dependencies or build steps. Serve this directory with any static HTTP server. For example, with Python installed:
python -m http.server 8000 --bind 127.0.0.1Open http://127.0.0.1:8000/. Check desktop and mobile layouts, navigation, image loading, project links and Escape-to-close behaviour.
Run the dependency-free checks with node --test services.test.mjs theme.test.cjs. They cover theme state and persistence, storage fallback, offer boundaries, safe default selections, critical-failure precedence, example arithmetic, priority filters and brief generation. Browser checks also cover interaction state, text rendering, copied output, screenshots and responsive layout. These checks are not a security or accessibility certification.
GitHub Pages serves the main branch at the repository root. .nojekyll keeps the authored assets as static files. Upload only these twenty reviewed files: index.html, styles.css, dark.css, theme.js, theme.test.cjs, site.js, services.mjs, services.test.mjs, fieldwork.png, planning-lab.png, fieldnote.png, fieldnote-mobile.png, .nojekyll, .gitignore, fieldwork-walkthrough.html, fieldwork-walkthrough.mp4, fieldwork-walkthrough-poster.jpg, Chatbot_QA_Sample_Report.pdf, Chatbot_QA_Sample_Scorecard.xlsx and this README.
No analytics, form submissions, external fonts or paid services are implemented. The only stored preference is dhairya-portfolio-theme in localStorage. The brief's text remains in page memory until navigation/reload or reset; explicit copy writes to the user's clipboard and explicit download saves a local file. Clipboard failure exposes a manual-copy fallback. The host can log connection information; external links follow their destination's privacy practices. No private phone number, email address, home address or résumé is published. Client material must not be entered into examples; permission and an appropriate processing arrangement are required for real work.
AI-assisted drafting and implementation are disclosed. No open-source license has been selected; public visibility alone does not grant reuse rights.
The illustrated walkthrough uses genuine project screenshots, labeled explanatory graphics, synthetic narration and burned-in English captions. It has native playback controls and an expandable transcript. It does not autoplay or add analytics. The examples remain fictional; the video makes no production integration or client-results claim.