Iksha is an independent effort towards mechanistic interpretability.

The lab's main focus is pre-intervention analysis of neural networks, making internal computation visible and accessible enough that researchers can ask better questions before applying causal tools.

We are NOT a product company, model provider, or a general MLOps platform.

COMING SOON.