Case studies
From decisions to delivery.
A selection of data engineering and analytics engagements across manufacturing, healthcare, energy, insurance, logistics, and IT services. Client names are withheld; the problems and the work are real.
Analytics platform for ServiceNow
IT service management
The problem
The client leaned on ServiceNow's native reporting, which could not handle complex groupings, custom slicing, multi-year trends, or custom KPIs without heavy Jelly and JavaScript to maintain.
What I did
I built a metadata-driven analytics platform on Azure. ETL ingested and transformed ServiceNow data into a data lake, a Data Vault model stored it cleanly, and a dimensional model on top served reporting.
The result
Custom KPIs and far better slicing in Power BI, an automated pipeline and refresh, low maintenance and operating cost, and a pattern that scaled to other ServiceNow entities.
Unified analytics for digital services infrastructure
Enterprise IT and digital services
The problem
Reporting was fragmented across O365, Citrix, SolarWinds, ThousandEyes, and more, each in a different format, so a single view of infrastructure health was nearly impossible.
What I did
I built a central metadata database for all sources, used a self-hosted integration runtime in ADF for on-prem data and a Databricks pipeline for API sources, and generated Data Vault tables from Jinja templates with a dimensional model for reporting.
The result
One unified view across every tool, automated ingestion and refresh, reusable metadata-driven code, and low operating cost.
Analytics and spend prediction for a utility
US energy provider
The problem
The client depended on a third-party vendor for analytics, which meant limited control and traceability, slow turnaround, data quality issues, and no visibility into usage patterns for marketing or service.
What I did
I built a secure Azure analytics platform with data quality, governance, and automation, added spend and usage prediction across residential, industrial, and government customers spanning 37+ customer types, and fed customer and invoice data into Salesforce.
The result
About $700K in yearly analytics savings, faster execution, real control and visibility over data, lower operating cost, and better customer service.
Golden-record MDM for a services firm
Global digital services company
The problem
Data was inconsistent, unvalidated, and unstandardized across enterprise systems, with no single trusted source for reporting.
What I did
I built an MDM solution on Databricks to create a golden source: consolidating entity versions across systems, then validating, cleansing, standardizing, and deduplicating into a 360-degree view.
The result
Attribute-level accuracy for key fields rose from about 93% to 98.8%, BI reconciliation tickets fell 38%, on a scalable, secure, cost-effective platform.
Reusable data framework on AWS
Enterprise data platform
The problem
The client needed a reusable framework to collect, store, catalog, and serve data faster and more securely. Slow ingestion, inflexible storage, and inefficient SSIS ETL held them back.
What I did
I used AWS DMS for real-time ingestion from SQL Server, implemented CDC into Redshift, shifted SSIS ETL to an ELT framework on Redshift's MPP engine, and orchestrated the transformations with Step Functions.
The result
Real-time insights, more efficient processing, automated workflows, cost-effective scaling, and secure, fast access to curated data.
Commercial analytics lakehouse
Commercial products manufacturer
The problem
The client needed a scalable foundation for behavioral analytics and predictive models, and wanted to explore GenAI-driven knowledge discovery.
What I did
I designed an Azure lakehouse with curated Delta tables and a reusable semantic layer, ran a proof of concept on vector databases and RAG, and built embedding and indexing workflows over curated Delta data.
The result
Governed reporting grew from 21 to 28 standardized KPIs, median dashboard latency dropped from 12.4s to 7.3s with Delta optimization and Z-ordering, and monthly compute fell 18 to 22%.
Teradata to Redshift migration for an airline
Major US airline
The problem
Analytics ran on Teradata, and the business needed to migrate to Amazon Redshift without losing performance.
What I did
I led the migration using AWS Schema Conversion Tool to automate DDL conversion, replaced Teradata stored procedures and OLAP functions with Redshift SQL, Lambda, and Step Functions, and tuned distribution and sort keys, WLM, and columnar storage.
The result
Analytics moved off Teradata onto a tuned Redshift platform, with performance preserved through careful key and workload management.