About MWirelabs
The Research & Advancement Wing of Mwire
Bridging the Digital Divide, One Language at a Time
MWirelabs was born from a simple realization: the world's most advanced AI systems couldn't understand a word of Garo, Pnar or Wancho?
In 2019, MWire began as a technology solutions provider, building digital systems for government departments and organizations across Meghalaya.
As we worked closely with communities across Northeast India, we saw firsthand how language became a barrier, not just to communication, but to participation in the digital economy itself.
By 2024, we made a strategic pivot. We established MWirelabs as our dedicated AI research division with a clear mission: build language AI from scratch for the 45+ million people across Northeast India whose languages have been ignored by global tech companies.
Today, we're among the few organization in India training foundational language models specifically for Northeast Indian languages, from generative AI to multilingual embeddings to classical NLP systems.
Our models power government services, support researchers worldwide, and prove that inclusive AI isn't just possible; it's essential.
What we stand for!
Our Mission
To advance open research in AI with a focus on inclusivity, safety, and cultural context. We build the complete AI stack for Northeast India: language models, speech systems, vision AI, and deployment-ready tools; ensuring technology serves every community in their own voice.
Our Mission
A future where language is never a barrier to accessing technology, information, or opportunity. Where AI understands not just words, but cultural context. Where communities from Meghalaya to Arunachal Pradesh can interact with technology as naturally as they speak with their neighbors.
Our Focus Areas
Low-Resource NLP
Creating foundational models for languages with limited digital data.
Open-Source Resources
Sharing models, papers, and datasets to empower the community.
Ethical & Responsible AI
Ensuring our technology is developed and deployed with a focus on fairness and safety.
Our Foundation Models
We create and open-source foundational models designed to support research, education, and practical applications.
Kren v1
Generative Mode
Kren v1 is a publicly documented encoder→decoder conversion producing a generative language model for an Indian language (Khasi).
~110M params
KhasiBERT
Language Model
Foundational Khasi model trained on ~3.6M sentences. Useful for translation, summarization, and low-resource NLP research.
~110 params Encoder
Ready to Collaborate?
MWirelabs is proud to drive research and innovation. If you have a project in mind, let’s connect.