Government Reporting Overhaul for a Home Care Platform
CorrDyn stabilized 100 mission-critical CMS reports for a home care company ahead of a hard February deadline, then transitioned to ongoing data team support.

~100 CMS-required reports
Government reports stabilized
February 1 CMS filing
Regulatory deadline met
DTaaS model post-launch
Ongoing data team
The situation
A home care coordination company that helps elderly people remain in their homes — connecting them with grocery delivery, transportation, and in-home nursing — had acquired another company with active government contracts and Medicare reporting requirements. The acquisition brought data obligations the company was not yet equipped to meet.
The data environment reflected the history: GCP and Azure both in use, transformation scripts scattered across Azure Data Factory, queries managed outside any version-controlled transformation layer, and reporting logic that no single person fully understood. Tableau dashboards sat downstream of this infrastructure, serving the analysts who needed to verify outputs before federal submission.
The hard date was February 1. CMS required approximately 100 mission-critical government reports. Missing the deadline was not an option.
What we built
CorrDyn’s first priority was the Azure environment, where the most urgent CMS reporting lived. We implemented dbt as the transformation layer, migrating the existing partner invoice logic and ad hoc scripts into governed, testable models. This gave the team version control and a clear audit trail for every transformation feeding into a government report.
We then built standardized entity models for the three core data objects the reporting touched: member, assessment, and cases. These shared definitions meant that different reports referencing the same concept were pulling from the same underlying model, not from independently maintained query logic that could drift out of sync.
The 837 EDI claim data, which carries Medicare billing transactions, required its own dbt review pass to confirm that the transformation logic matched CMS specification. We worked through that alongside onboarding the client’s internal analysts so they understood the new model structure and could maintain it independently over time.
Planning for the post-deadline work also began during this phase: implementing Estuary for change data capture from SQL Server into the transformation layer, and addressing the GCP/BigQuery side of the environment that the initial engagement had set aside.
What changed
The February 1 CMS filing deadline was met. The approximately 100 government reports that had been running on fragile, undocumented logic were moved onto a governed dbt layer with tests and documented lineage. Analysts could trace any number back to its source.
After the deadline, the engagement transitioned to a managed data team model at a fixed monthly budget. CorrDyn continues to handle dbt model development, new reporting requirements, and infrastructure work as the company builds out its multi-cloud data environment.
Frequently Asked
Questions
We inherited broken data infrastructure through an acquisition and have a regulatory deadline — can CorrDyn move fast enough?
How does CorrDyn handle multi-cloud environments with data split across GCP and Azure?
After the crisis is resolved, do we need to hire a permanent data team?
Related Work
Similar engagements across our portfolio.

Data Platform Migration for a Healthcare Logistics Company
CorrDyn migrated a healthcare logistics company from Snowflake to MotherDuck in 8 days, then transitioned to ongoing data team support for plan reporting.

Embedded Analytics Modernization for a Payroll SaaS Platform
CorrDyn built an embedded analytics PoC for a payroll platform, achieving sub-second latency and 6-minute data freshness at 75% lower cost.

ML Customer Segmentation for an Automotive Dealer Group
CorrDyn built an ML clustering pipeline on Databricks with LLM-powered profiling to segment customers across vehicle brands for a dealer group.
Related Conversations
Podcast episodes that cover the same ground.
Unlocking the Power of Semantic BI with Hashboard
with Carlos Aguilar, Hashboard
Carlos Aguilar explains how Hashboard eliminates multiple sources of truth in analytics through semantic BI, version control, and governance.
From Tissue to Mechanism to Decision: AI for Computational Oncology
with Arvind Rao, University of Michigan
Arvind Rao of Michigan on SPIFEE and the tumor microenvironment, multimodal AI from tissue to drug target, and why clinical AI needs credentialing.
Cavities in the Data: Building FDA-Cleared AI for Dental Imaging
with Sadegh Salehi, Overjet
Overjet's Sadegh Salehi on subgroup evaluation, sensor shortcuts in training data, and the FDA's Predetermined Change Control Plan for medical AI.