-
Notifications
You must be signed in to change notification settings - Fork 0
Home
Welcome to the official technical documentation for the Payer Sanskritkurs Modernization & Translation Project.
Note
This project digitizes and modernizes the renowned Sanskrit course by Prof. Dr. Alois Payer, transforming legacy HTML manuscripts into an ultra-fast, multi-lingual VitePress Progressive Web App (PWA) supported by local AI translation models.
| Pillar | Technology | Goal |
|---|---|---|
| πͺΆ Scholarly Heritage | VitePress, DevanΔgarΔ« & IAST standards | Preserve typographical precision & academic rigor |
| π€ AI Translation | Weg B (Qwen3.6-35B-A3B on nyx.local) |
Zero-cost, 100% clean translations across 35 locales |
| βΈοΈ CI/CD & Staging | GitHub Runner + Docker on nataraja.local
|
32GB RAM automated builds, local live-staging & vector search |
| π‘οΈ Code Governance | Regular Code Reviews via Claude Opus | Continuous architectural audits, QA gate verification & zero-defect pipeline |
Comprehensive design guide covering our 3-node hardware split (Workstation: nike.local, nyx.local LLM Engine, and nataraja.local Linux Server), Mermaid system diagrams, CI/CD staging pipeline, Claude Opus code review governance, and historical change logs.
Detailed specification of our 3-stage LLM translation hierarchy (Local 35B model, Claude Sonnet, Gemini 2.5 Pro), DevanΔgarΔ« protection filters, TM cache management, and quality control routines.
Historical background of the original manuscript, the transition from legacy 1990s HTML to modern VitePress PWA, and internationalization goals.
Tip
Use the Sidebar navigation on the right to jump directly to any topic or check live runner and staging endpoints.
Payer Sanskritkurs Wiki β’ Modern High-Performance AI Translation & Publishing System β’ The Illuminated Manuscript of the Future
- π Home
- π System Architecture
- π€ AI Translation Methods
- π« Language Selection & Exclusion Criteria
- ποΈ Prof. Payer's Sanskrit Course
- π» nike.local: Primary Mac (M2, 24GB VRAM)
- π nyx.local: MacBook Air M4 (32GB VRAM)
-
Qwen3.6-35B-A3B(~20 t/s)
-
- βΈοΈ nataraja.local: Intel i7 (32GB RAM, Pop!_OS)
- π Staging:
http://nataraja.local:8080 - π§ Ollama:
http://nataraja.local:11434 - π Vault Archiver
- π Staging:
- π GitHub Repository
- π¦ Container Registry
- β‘ CI/CD Runs