Sanjeev Azad
← All work

Open source

Data Darśana

A living map of the data architecture behind analytics, generative AI, and agents: 248 capabilities and 128 patterns as one navigable territory.

1B6F · packing

No one designs a data estate. It accumulates.

A warehouse from one era, a lake from the next, a mesh someone announced and half-built, and now a scramble of vector indexes and agents bolted to the side. Every layer was a reasonable decision at the time. Together they are a landscape nobody can hold in their head — and you are asked to draw a target architecture for it by Thursday.

Data Darśana is a map for that moment. Not a maturity model that scores you, and not a reference architecture that assumes you are starting from nothing. A map: the whole territory laid out at once, so you can find where you actually stand, see what genuinely connects to what, and make the next decision with your eyes open.

The gap it closes

Most capability models tell you what you must be able to do. Reference architectures tell you what good looks like in one shape. Almost nothing connects the two — and that gap is precisely where architecture lives, where the arguments happen, and where budgets are won or lost.

Data Darśana holds both halves and joins them:

  • Capabilities: what must we be able to do? 248 capabilities across 20 domains and 6 pillars, each placed at the stage of maturity where it first genuinely matters.
  • Archetypes: how do we actually assemble it? 128 architecture patterns in 13 families, each carrying when to use it and what it costs you.
Capabilities Pillars Domains Archetypes Families Links Diagrams
248 6 20 128 13 1,888 112

Try it

MIT licensed.