About this site.
This website has two ways of being read: one composed for people, and one structured for machines. They are not separate accounts. They are views of the same public record.
The premise
Information should be able to change form without losing its meaning, provenance or accountability.
01 · Human view
Written to be understood.
The Human view arranges the work as narrative. It provides emphasis, sequence and context: what I do, what I have built, where I have worked and why the work matters.
It is deliberately readable without exposing the machinery underneath every sentence.
02 · Machine view
Structured to be interpreted.
The Machine view is a human-readable inspector for the structured data behind the site. It exposes the shape of the record—its entities, relationships, time, claims, evidence and provenance—in a form a person can examine.
Machines do not need to read that visual interface. They can access the canonical JSON record directly. The inspector exists so that people can see what the machine-readable version says.
03 · One record
One source of truth.
The two views are generated from one manually maintained canonical record. The record identifies people, organisations, roles, work, credentials, recognition and activities; it also describes how those things relate to one another.
The smaller profiles and page-specific structured data are projections, not independently maintained copies. A change is made once, validated, and then propagated into the forms that need it.
04 · Evidence + change
Claims that can show their workings.
Selected public assertions are connected to evidence at the claim level. A source must support the specific assertion—not merely mention its subject. Internal or person-attested information is identified differently from externally verified information.
The current record has a stable address at https://sannuta.com/sannuta-raghu.record.json. Established versions are retained as immutable snapshots, so later corrections or additions do not silently rewrite the earlier public record.
05 · Making it
Human-directed. Machine-assisted.
The concept, language, information architecture and editorial decisions are mine. I decided what the record should contain, how claims should be expressed, what required evidence and what the interface should communicate.
The site was developed iteratively with OpenAI Codex, primarily using 5.6 Sol. The system helped translate those decisions into HTML, CSS, JavaScript, JSON Schema, migration scripts, validators and deployment tooling. Every substantive change was reviewed and approved by me.
The site is a small demonstration of its own argument: factual information can become fluid without becoming unmoored.
Return to the work ↗