Advancing scientific intelligence for the physical world
Foundation models, scientific machine learning and computational systems for engineering and industry.
At a glance
Dhīr base models
2
Technical team
65+
Publication
1
TPU v6e chips in production
2,048
As of October 2026
Featured research
Compact foundation models
Dhīr
Dhīr 2B and Dhīr 5B MoE are compact foundation models pretrained from scratch in India, for offline use on laptops, embedded boards and edge devices, in English and ten Indian languages.
Scientific foundation models: computing physics, not only describing it
Describing the Navier–Stokes equations is not the same as working with them. Our scientific foundation models, starting with Dhīr 9B, are being trained to interpret simulation outputs, check equation formulations and work through derivations that respect physical laws.
DescribesDescription only
“The outcome depends on several factors, including the materials, the geometry and the operating conditions. Exact values would require further analysis.”
Plausible, but nothing to measure
ComputesComputed field
u = …Illustration computed by a reference solver in your browser; not output of a Dhīr model.
Research is built by scientists and engineers
Founder & Chief Scientist
Prof. Dr. Sashikumaar Ganesan
Sashikumaar Ganesan is the founder and Chief Scientist of ZenteiQ, where he sets the company’s scientific direction. His work spans scientific computing, high-performance computing and scientific machine learning. That foundation shapes how ZenteiQ builds AI from the ground up, and its long-term focus on Scientific Intelligence and Physical AI.
Technical team
Model research, data and engineering work together in one team, from training runs to deployed systems.
Research scientists
AI engineers
Data engineers
Platform engineers
FMOps engineers
Full-stack engineers
Research context & ecosystem
IndiaAI Mission
Dhīr model development is supported under the IndiaAI Mission’s programme for foundation models.
ARTPARK
Joint work on the pipelines that prepare scientific and engineering data for model training.
The AI Alliance
Participation in open-science work on foundation models and open tooling.
Discuss the science behind our models
Technical architecture, evaluation approach and benchmark materials, shared with qualified partners under NDA.
