Built an OSS toolkit to normalize mixed-script/mixed-language user text before LLM pipelines.
Focus is production use: API mode, Docker deploy path, language-pack interface, and evaluation snapshots.
Would appreciate feedback on data strategy, evaluation design, and integration patterns.
https://github.com/SudhirGadhvi/open-vernacular-ai-kit
Show HN: Open Vernacular AI Kit – preprocessing for Indian code-mixed text | Heykuki News