Morphological Research
Advancing NLP architectures structurally tailored to the unique syntax, orthography, and morphology of Meghalaya's indigenous languages.
Like the living root bridges of our hills, language infrastructure grows stronger when woven by the community. We research, curate, and deploy foundational AI for Khasi, Pnar, Garo, and regional dialects.
Low-resource should never mean second-class engineering. We treat indigenous languages with the rigor of frontier machine learning while protecting cultural context.
Advancing NLP architectures structurally tailored to the unique syntax, orthography, and morphology of Meghalaya's indigenous languages.
Building high-purity, gated parallel datasets and standardized evaluation benchmarks so regional models are trained on verified ground truth.
Moving models out of academic notebooks and into fast, accessible web and mobile interfaces that work reliably in real-world conditions.
Enforcing transparent, community-first AI development that explicitly respects local ownership and cultural nuance over blind scraping.
How we systematically digitize and elevate Meghalaya’s linguistic heritage for the modern computational era.
Developing bidirectional NMT pipelines and custom tokenizers optimized for Khasi, Pnar, Garo, and related regional variations—bridging local knowledge with global accessibility.
Preserving endangered print literature, historical texts, and oral histories through custom Optical Character Recognition (OCR), corpus curation, and community annotation tools.
Engineering multimodal speech synthesis (TTS/ASR) and conversational assistants that genuinely understand local idioms, prosody, and Meghalayan cultural context.
Speak it · Translate it · Understand it
Natural-sounding voices and accurate transcription, built for local pronunciation, rhythm, and accents.
Break language barriers with translation systems designed around the languages and context of Meghalaya.
A local AI assistant that understands Meghalaya's languages, culture, and context — making AI accessible to everyone.
HOSTED ON HUGGING FACE & GITHUB // APACHE-2.0
NMT // ARCH
Spearheading Neural Machine Translation architectures, dataset curation, and full-stack deployment for low-resource languages.
ASR // OCR
Driving Optical Character Recognition development for low-resource scripts, alongside ASR and TTS synthesis for Khasic languages.