Democratise AI
Lower the cost to build and deploy AI in India. Support developers with open models and tooling that run on modest compute and at the edge.
Our goal is to develop AI for India's needs and to help democratise AI across Bharat. We work on systems that are practical, safe, and inclusive for a billion users.
Kashi Labs is an India first AI research lab. We aim to advance open AI research, build deployable technologies for public digital infrastructure, and enable startups, developers, and institutions to adopt safe machine intelligence.
Lower the cost to build and deploy AI in India. Support developers with open models and tooling that run on modest compute and at the edge.
Design for India's diversity, culture, and heritage. Build robust systems that work across scripts, bandwidths, and devices.
Align models with Indian law and context. Privacy preserving methods, auditability, and real world red teaming.
Text, code, and math-native transformer models. Instruction-tuned text-to-text systems for reasoning, coding assistance, and developer agents.
Text-to-image and text-to-video generation for marketing, advertising, and storytelling. Controllable, safe media pipelines adapted to Indian languages and contexts.
CNNs, RL, and GANs for medical imaging, factory automation, and optimized supply-chain logistics. High-reliability models designed for real-world operations.
UPI, ONDC, ABHA, DigiLocker. Trustworthy agents and middleware that respect user consent and Indian regulations.
Benchmarks for Indian languages and scenarios. Red team methods, watermarking, and policy toolkits for responsible AI.
Datasets, model hubs, enterprise agent management, and SDKs to accelerate startups and researchers, supported by practical documentation and examples.