A source-preserving research reader, legacy wiki modernization, and tools for working with a structured corpus.
My Vani-related work spans product interfaces, content ingestion, MediaWiki compatibility, deployment tooling and operational investigation. This repository brings the engineering decisions and evidence together.
Live Vedawiki reader · Public collection coverage · My profile
| Project | Engineering focus | Status and detail |
|---|---|---|
| Vedawiki reader | TypeScript/Next.js, source metadata, indexed search, resumable imports, accessible reading interactions and Azure releases | Live. Reader case study |
| MediaWiki modernization | An eight-wiki Docker migration environment, staged database upgrades, PHP extension/skin compatibility, selective deployment and rollback scripts | Implementation and migration reports available privately; production cutover is not asserted. Migration case study |
| Veda MCP tooling | Typed corpus records, exact citation lookup, bounded search, structured tool responses and explicit content-rights metadata | Local tooling/prototype. Corpus and MCP design |
| Operational assessment | Read-only Linux investigation, backup evidence, privilege boundaries and separating observations from conclusions | Assessment methodology. Operations notes |
On 20 October 2026, Vedawiki's public health and coverage endpoints reported 1,551 imported source pages, 0 failed jobs and 0 pending jobs in snapshot db939f6091395225e05b.
These counts cover five named collections and selected supporting pages. They are not the size of the entire Vani archive, unique verse counts, user numbers or an uptime guarantee. The snapshot was generated on 18 October 2026. Recorded evidence and its limits.
flowchart LR
A["Original Vani editorial sources"] --> B["Bounded import and source metadata"]
B --> C["Versioned corpus and search index"]
C --> D["Vedawiki reader on Azure"]
D --> E["Reading, search, citations and local bookmarks"]
A -. "Explicit live fallback" .-> D
F["Local corpus adapters"] --> G["Typed records with rights metadata"]
G --> H["Veda MCP lookup and search tools"]
The reader and the MCP prototype are separate implementations. The diagram does not imply that the public reader exposes an MCP endpoint or that the prototype uses the reader's production snapshot.
The small helper below was added for this showcase. It reads only the public health and coverage endpoints, checks snapshot/count consistency, and prints a dated summary.
python3 scripts/verify_public_reader.pyPython 3.10+; standard library only. It does not log in, crawl the collection, refresh source pages or change the service.
This is a curated engineering showcase, not a release of the private production repositories. The migration work adapts an existing platform and team-maintained customizations; it does not claim authorship of MediaWiki, Semantic MediaWiki, all Vani extensions, or the source texts. AI-assisted implementation and debugging are part of my workflow.
The public material contains fresh case-study documentation, a public-endpoint verification helper and a small metadata snapshot. Operational configuration, database exports, private logs, audit findings and source content are not distributed here. Writings and media remain with their respective rights holders.