The doctoral research behind AlifZetta — Padam Sundar Kafle on SI
Twenty-one years of engineering meets a doctoral thesis on Superintelligence.
The engineer arc
Padam Sundar Kafle has shipped software for twenty-one years, across thirty-five countries, in healthcare (HTE), wearables (ZettaBand), video analytics (IrisVision), and now sovereign SI (AlifZetta). Systems that survive in the field, not just on the whiteboard.
The doctoral thesis
The current doctoral research asks: is Superintelligence a training-time property or a substrate-time property? AlifZetta is the falsifiable answer — a substrate that grows every thirty minutes and reasons over what it just learned, without a retrain.
Why the founder-as-KB matters
Padam ships his own bio as a first-class NEXUS entity — the founder KB. The platform (AlifZetta) is documented in full. If you cannot cite yourself, you cannot ask your users to cite you.
Why we use different words
Where the industry says LLM, we say CLLM. Where it says vector database, we say NEXUS. Where it says Chain-of-Thought, we say PRISM proof-tree. Where it says RAG, we say LATTICE. Where it says trillion-token dataset, we say Smart Router Dataset. Different words because different architecture. Read the whitepaper.