NomadLake
Bronze. Silver. Gold. One source of truth.
NomadLake consolidates every subsidiary across a conglomerate into one medallion lakehouse. Raw records from the whole group land in immutable bronze, get cleaned and identity-resolved into silver, and roll up into the gold marts your board reads. You read finished, trustworthy group numbers, trace every one back to the source system, and resolve one customer across every sector.
Not a video.The real app, live. No sign-up.
Live demo on a deterministic sample conglomerate across ten sectors.
NomadLake. Every system, one source of truth
By the numbers
- medallion layers: bronze, silver, gold
- 3
- sector schemas, plus a shared party 360
- 10
- data-quality dimensions scored
- 4
- every gold figure traced to source
- Full lineage
medallion layers: bronze, silver, gold
sector schemas, plus a shared party 360
data-quality dimensions scored
every gold figure traced to source
How a number is made
Bronze in, gold out, every step on record
Raw events land, get cleaned, and roll up into the trusted numbers your dashboards read. You can trace any one of them back.
- 1
Bronze
Raw events from every system, kept exactly as they landed, on write-once storage.
Raw in, untouched
- 2
Silver
Cleaned and conformed: identities resolved through the party 360 graph, scored on completeness, accuracy, timeliness, and consistency.
Cleaned and scored
- 3
Gold
Analytics-ready marts and materialized views your dashboards read, with revenue trends and full lineage back to source.
Trusted facts out
Capabilities
What it does
- One source of truth across the group
- Core banking, insurance, distribution, manufacturing, and the rest of the group, 50 to 70 source systems across ten sectors, land in one place and reconcile into a single trustworthy dataset.
- No data team required
- The pipeline cleans and aggregates on a schedule. You read finished numbers, not raw exports.
- Trace every number
- Data lineage traces every gold figure back to the system it came from.
- Data quality scores
- Trust scores against target across four data-quality dimensions.
- Azure lakehouse, free local twin
- Runs as an Azure Databricks lakehouse on immutable storage in production, and the same code runs free on a local open-source twin, so a team builds the real thing before going live.
How it is built
Every subsidiary in, one trustworthy set of group numbers out
A conglomerate runs 50 to 70 source systems across a dozen sectors. NomadLake lands them all in an immutable Azure lakehouse, resolves one customer across every business, scores and traces every figure, and serves finished group numbers a board can defend.
Source systems
50 to 70 across 10 sectors
- Bank and merchant bankRegulated core, change-data-capture
- Insurance and pensionPolicy, claims, members
- Distribution and autoERP and dealership systems
- Beverage and packagingProduction and route delivery
- Real estate, media, utilitiesThe remaining sectors
Azure lakehouse
Bronze, silver, gold
ADLS Gen2Immutable WORM bronze
DatabricksIceberg, Unity Catalog- Quality gatesPer-sector checks
Resolved and traced
Trusted by construction
- Party 360One customer across 10 sectors
- Purview lineageEvery figure to source
- AML complianceSTR filing, 6-year retention
Group reporting
Finished numbers
- Conglomerate dashboardsGroup and per sector
- Power BI and Databricks SQLThe serving layer
The same code runs free on a local open-source stack and on Azure with one env-var swap, so local development is a real dress rehearsal, not a mock. The public demo runs on a deterministic sample conglomerate.
The stack
An Azure lakehouse, with a free local twin
The production home is Azure: a Databricks lakehouse on immutable storage, governed by Microsoft Purview, behind private networking. The same code runs locally on open-source twins, so a team builds the real thing for free and goes live with an env-var swap.
- Lakehouse on Azure
- Azure DatabricksADLS Gen2Unity Catalog
Apache Iceberg - Ingest and transform
- Azure Data FactoryEvent Hubs + Debezium
dbt - Govern, resolve, serve
- Microsoft PurviewSplink resolutionPower BI
- The same stack locally, free
Postgres
Trino
DagsterMinIO and Soda
Who it helps
One platform, three jobs done
The same lakehouse serves the people who build the numbers, the people who decide on them, and the people who have to prove them.
Data team and analysts
The people who build the numbers
- One pipeline cleans and scores every source on a schedule, so you publish finished numbers instead of hand-built exports.
- Trace any gold figure back to the exact source record through the lineage catalog.
- Quality checks gate every layer, so a bad feed stops before it reaches a dashboard.
Leaders and executives
The people who decide on them
- One trustworthy set of numbers across every business, not a different total per system.
- See the group and each sector side by side, with revenue trends already rolled up.
- One customer resolved across every sector, so you know who you actually deal with.
Compliance and audit
The people who have to prove them
- Immutable, write-once bronze means every figure traces back to an unedited source record.
- Reproduce any number as of a past date from table snapshots.
- Subject-access requests resolve against one identity graph, not a search per system.
Architecture and security
Governed and audit-ready
- Write-once raw layer.
- Bronze lands on object-lock storage that cannot be edited or deleted. Bank-grade immutability for an audit.
- Time-travel and lineage.
- Table snapshots let you reproduce any figure as of a past date; a lineage catalog traces it from gold mart back to the bronze file.
- One customer across systems.
- A two-tier identity graph resolves the same party across all ten sectors, so a number ties back to one entity, not a duplicate per source.
- Private and identity-bound.
- It runs on private endpoints with Managed Identity to storage. No storage keys in code, no public surface.
- Quality gates the pipeline.
- Data-quality checks run at every layer; a failed check stops bad data before it reaches a dashboard.
- Azure-native, free to build locally.
- Runs as an Azure Databricks lakehouse on ADLS Gen2 in production; the same code runs free on a local open-source twin, so local development is a real dress rehearsal and going live is an env-var swap.
Try it
What you can try
Five minutes, no sign-up required.
- 1Walk the bronze, silver, and gold layers and see how raw events become finished numbers
- 2Trace a gold figure back to the source transaction it came from
- 3Check the trust scores across the four data-quality dimensions
- 4Browse the breakdown across the ten industry schemas and open the party 360 identity graph
Under the hood
How it works.
Want software built around your business?
Tell us the goal in your own words. We reply with a 60-minute call, an honest yes-or-no on whether we are the right fit, and a one-page sketch of how it could work. No obligation.
- Replies within one working day, usually sooner. Trinidad business hours (UTC-4).
- Your details stay between us. See the privacy notice.