We build India's AI stack end to end, with partners at every layer: data, research, models and deployments.
Much of India is not on the internet, so our teams went to the ground.
Data sourced across India's states and union territories, through our own points of contact.
Text from local publications and other sources that are not on the internet, led by Prof. Aditya Maheshwari's team at IIM Indore.
Audio recorded across Indian languages, so our speech models hear India the way it speaks.
Our linguists check data and models across 22 Indian languages, against benchmarks built for India, not translated.
We work with publishers and archives on terms fair to both sides. If you hold Indian-language text, speech or documents, let's talk.
Nine of India's top academic institutions, anchored at IIT Bombay. Their research flows straight into our models.
Anchors the consortium, with the largest group of researchers, working across foundation AI models, efficiency, NLP and speech.
LLMs · Efficient AI · NLP
Prof. Ganesh Ramakrishnan, Dr. Aditya Desai, Dr. Pratik Jawanpuria, Dr. Preethi Jyothi, Dr. Arpit Agarwal
Vision research at flagship conferences, flowing straight into Patram.
LLMs · NLP · Vision
Prof. Ravi Kiran Sarvadevabhatla, Prof. Priyesh Shukla, Prof. Hrishikesh Terdalkar
Quantum-inspired lightweight models, dialects and manuscripts.
Specialised LLMs · Manuscripts · Quantum
Prof. Rohit Saluja, Prof. Abhishek Sharma
Linguistic grounding, legal models in Indian languages and grammar correction.
NLP · Specialised LLMs · Manuscripts
Prof. Arnab Bhattacharya
Indian-language-aware tokenisation and cultural knowledge in LLMs.
LLMs · NLP · Manuscripts
Prof. Maunendra Desarkar
Ground-level data collection and the IndicParam benchmark, accepted at EMNLP 2026.
Efficient AI · Specialised LLMs · Manuscripts
Prof. Aditya Maheshwari
Research on foundation AI models, efficient training and alignment.
LLMs · Efficient AI · Alignment
Prof. Bapi Chatterjee
Research on speech and on aligning models after training.
Speech · Alignment
Prof. Madhusudhanan
Founding member of the BharatGen consortium.
22 researchers from 14 institutions, including IISc, IISER Kolkata, BITS Pilani, Central University of Punjab, IIT Ropar and LNMIIT. Pick an area to see who works on it.
BITS Pilani
Central University of Punjab
IIIT Delhi
IIM Indore
IISER Kolkata
IIT Hyderabad
IIT Kanpur
IIT Madras
IIT Ropar
LNMIIT
Collaborations that bring our work to the global research community.
JapanA three-way agreement between IIT Bombay, BharatGen and NII on language models with stronger scientific reasoning, with joint research, researcher exchanges and shared datasets.
AustraliaA three-way agreement between IIT Bombay, BharatGen and Monash on trusted infrastructure and culturally inclusive AI.
GlobalBharatGen is a founding contributor to Project Tapestry, which helps nations pool data, compute and talent while each keeps control of its own models and data.
Yann LeCun welcomed us to Tapestry ↗
About Project Tapestry →BharatGen is one of 60 organisations across Asia, Africa, Europe and the Americas to sign Ensuring AI speaks everyone's language, a joint commitment led by the Gates Foundation. AI labs, governments, researchers and community groups share one goal: that within five years, 3.4 billion people whose languages AI underserves can use it in their own language and voice.
The compute, cloud and infrastructure partners that help us train and serve our models.
Organisations using our models in their own work.
Three ways to work with BharatGen.
For developers and enterprises with in-house AI teams. Download the open weights and build your own Indian-language applications.
For organisations that want hands-on support. We scope, build and deploy a solution with you, on your infrastructure.
For enterprises that prefer an existing technology partner. IT services firms deliver solutions built on BharatGen's models.
Tell us what you are working on, whether you have data, research or a problem to solve. We read every message.