In the field.
Two engagements, both live. One running in production; one in active build-out. Real systems, not mockups. Client names withheld out of respect for our partners — the work is real.
Property Manager
A multi-property residential operator running a full analytics pipeline for delinquency, occupancy, and renewal intelligence — built and maintained by us, with a daily-refreshed dashboard for the client's team.
Challenge
Delinquency tracking was lived in spreadsheets. Occupancy was reported monthly — long after the window to do anything about it had closed. Renewal risk was a gut feel, not a metric. The operator wanted decision-grade reporting without hiring a data team.
Approach
- Stood up a dedicated GCP project and BigQuery warehouse.
- Built a Dataform pipeline: raw → cleaned → dashboard-ready, version-controlled in GitHub.
- Deployed a Metabase dashboard scoped to the client's collection, with permissioned access.
- Used AI to draft SQL transforms and dashboard cards at speed — every change reviewed and shipped by a human on our team.
Outcome
- Daily refresh of delinquency, occupancy, and renewal metrics.
- Axis scaling tuned so meaningful variance actually shows up — not a flat line at 90%.
- New reports shipped in hours, not weeks — requests come in plain English, we turn them around.
- Live on our hosted Metabase instance, permissioned to the client login.
Protein Bar Company
Reporting stack and alert tooling for a protein bar brand — warehouse, transforms, and dashboards, currently in active build-out.
Challenge
Sales, inventory, and channel data scattered across platforms. No unified view of performance. No way to catch anomalies before they show up in a monthly summary — by which point the quarter is half over.
Approach
- BigQuery warehouse in a dedicated GCP project.
- Dataform-driven transformations with a strict dashboard layer — no card queries intermediate tables.
- Metabase collection and permissioning set up for internal stakeholders.
- AI-surfaced alerts on revenue and conversion anomalies (in design).
Status
Warehouse provisioned, initial ingest schemas defined, dashboard layer scaffolded. Anomaly alerting and client-facing dashboard build scheduled next.