Project description
Our data is scattered everywhere, just not in one place. Sales in CRM, orders in billing, ad spend in platform dashboards, product events in separate analytics, and finance keeps their own spreadsheets. Every time a simple question comes up — how much did we really earn from a specific campaign or which customers left last quarter — an analyst spends two days on exports, manually matches identifiers, and hopes the sums add up. We want to stop this. The goal of the project is a single cloud data warehouse, Snowflake or BigQuery, that becomes one trusted source for all queries.
We are looking for an engineer who will design and build the entire pipeline from start to finish. It starts with loading: each source needs to be brought into the warehouse on a reliable schedule — through ready-made connectors or through custom export where there is no connector — with careful handling of incremental loads and change history. On top of the raw layer, data should be properly organized using dbt: staging models that clean and standardize each source, intermediate models with joins and deduplication, and clear marts that speak the language of business. Tests, documentation, and version control are part of the deliverable, not something to be done later, so that transformations are transparent and can be safely changed.
The final part is a semantic layer with agreed-upon definitions of key metrics: revenue, active users, margin, retention — so the same number means the same thing in all reports. Next, we need to connect our BI tool to these prepared models and assemble several starter dashboards that prove the warehouse works and give the analytics team a foundation for further development. Please describe your approach to modeling, how you'll keep costs under control, and roughly how long the initial build will take.
— Scattered sources: CRM, billing, ad dashboards, product events, financial spreadsheets
— Warehouse: Snowflake or BigQuery, your recommendation with justification
— dbt models with tests, documentation, and version control
— Semantic layer with agreed metric definitions
— BI connection and several starter dashboards
We are looking for an engineer who will design and build the entire pipeline from start to finish. It starts with loading: each source needs to be brought into the warehouse on a reliable schedule — through ready-made connectors or through custom export where there is no connector — with careful handling of incremental loads and change history. On top of the raw layer, data should be properly organized using dbt: staging models that clean and standardize each source, intermediate models with joins and deduplication, and clear marts that speak the language of business. Tests, documentation, and version control are part of the deliverable, not something to be done later, so that transformations are transparent and can be safely changed.
The final part is a semantic layer with agreed-upon definitions of key metrics: revenue, active users, margin, retention — so the same number means the same thing in all reports. Next, we need to connect our BI tool to these prepared models and assemble several starter dashboards that prove the warehouse works and give the analytics team a foundation for further development. Please describe your approach to modeling, how you'll keep costs under control, and roughly how long the initial build will take.
— Scattered sources: CRM, billing, ad dashboards, product events, financial spreadsheets
— Warehouse: Snowflake or BigQuery, your recommendation with justification
— dbt models with tests, documentation, and version control
— Semantic layer with agreed metric definitions
— BI connection and several starter dashboards
Other / Miscellaneous Cash, Bank transfer
