Project description
We're a mid-sized company and over the years we've accumulated many databases, schemas in our data warehouse, exports in spreadsheets, and BI reports, but no one has a complete picture of what data exists and where it lives. Analysts spend days searching for the right table, ask colleagues in chats, and rebuild metrics that have already been calculated somewhere. When two reports disagree on numbers, we can't quickly figure out which source is correct and who owns it. We want to solve this with a proper data catalog: a single searchable registry of datasets, tables, columns, and reports, where each has a clear description, business meaning, owner, and freshness indicator.
Beyond search, we need lineage — the ability to see where a field comes from, which pipelines feed a table, and which dashboards will break if we change it. It's also important that data quality flags are visible right in the catalog: a person immediately understands if it's a trusted set, outdated, or under review. The main goal is trust: an analyst opens the catalog, finds the dataset they need in seconds, understands its meaning, sees who to contact, and already knows whether they can rely on the numbers before building anything on them.
We're willing to consider a suitable open-source or lightweight solution instead of a heavy enterprise platform if it fits our stack and can be maintained by a small team. We ask that you propose an approach and approximate timelines.
What we expect from the work:
— Inventory and automatic metadata collection from sources;
— Search with descriptions, owners, and tags;
— Lineage visualization at table and column level;
— Data quality indicators and a simple verification process.
Beyond search, we need lineage — the ability to see where a field comes from, which pipelines feed a table, and which dashboards will break if we change it. It's also important that data quality flags are visible right in the catalog: a person immediately understands if it's a trusted set, outdated, or under review. The main goal is trust: an analyst opens the catalog, finds the dataset they need in seconds, understands its meaning, sees who to contact, and already knows whether they can rely on the numbers before building anything on them.
We're willing to consider a suitable open-source or lightweight solution instead of a heavy enterprise platform if it fits our stack and can be maintained by a small team. We ask that you propose an approach and approximate timelines.
What we expect from the work:
— Inventory and automatic metadata collection from sources;
— Search with descriptions, owners, and tags;
— Lineage visualization at table and column level;
— Data quality indicators and a simple verification process.
Other / Miscellaneous Cash, Bank transfer
