Ten years building the data behind workforce decisions.
At Allegis Group I helped build the analytical warehouse for one of the largest US staffing organizations. At Magnit Global I lead the team that unifies data from multiple VMS platforms into one governed analytics layer for 110+ enterprise clients. The hard part was never the dashboards. It was one trusted definition, one identity per supplier and worker, and data leaders could act on.
-
2010 – 2016
CareFirst BlueCross BlueShield
Programmer Analyst II
Regulated, high-volume data: claims, member and provider data with PHI handled under HIPAA. Where I learned governance first.
-
2016 – 2019
Allegis Group
Senior Data Engineer / DW Architect · staffing
Allegis Analytical Warehouse: bus matrix and conformed dimensions across business units, Informatica ETL across 10+ source systems, MDM golden-record patterns.
-
2019 – today
Magnit Global
Engineering Manager, Data Analytics & AI · workforce solutions
Workforce analytics platform for 110+ clients across SAP Fieldglass, Beeline, VNDLY and other VMS sources. Snowflake live since May 2025; Power BI semantic layer with row-level security; AI on governed KPIs.
-
Next
Your workforce program
The same patterns, applied to your data
The working demo below and the 90-day plan show how I'd bring them to your organization.
Delivered What I've actually led
What workforce analytics needs, and where I've built it Delivered
| What the program needs | Allegis · staffing | Magnit Global · today |
|---|---|---|
| One definition of spend, headcount and rateTrusted Metrics Foundation | Conformed dimensions and fact tables across business units. | Lean Data Model: reusable business definitions and thin semantic models across 5+ VMS platforms, cutting new-client onboarding effort ~40%. |
| One identity per supplierSupplier master data | Informatica MDM matching and golden-record patterns. | AI-assisted supplier MDM: matching, hierarchy detection and metadata enrichment across VMS systems. |
| AI on governed metricsIdea · AI investigator | KPIs and analytical models defined with business stakeholders across divisions. | Snowflake Cortex pilot: analysts ask plain-English questions against governed KPI views, with accuracy evaluated before wider rollout. |
| Reliable, governed platformIngestion · security · release control | Informatica IDQ data quality; CI/CD for warehouse releases. | Snowflake Bronze/Silver/Gold with Kafka CDC, Fivetran, dbt and Airflow; Power BI with row-level security replacing per-client reports; Oracle → PostgreSQL migration of 100+ reports with zero client disruption. |
One decision loop for the whole workforce program
Every question a program leader asks fits the same four steps. Pick a stage to see the question it answers.
How I'd answer "why did contingent spend go up?"
A working demo of the approach, not a product pitch. Pick a question: each answer shows the governed metrics it used, the data checks it ran, and where a person has to decide. It builds on what I've already started with a Snowflake Cortex pilot over governed KPI views.
Baseline → intervention → outcome → attribution → learn. Synthetic figures; the method is the point.
One trusted number for headcount, spend and rate, in every tool.
Technically: a platform-independent semantic layer over every VMS
Each VMS counts workers, spend and rates its own way, and finance counts invoices. So a simple question gets five answers. This is the problem my Lean Data Model solves today across 5+ VMS platforms: define each metric once, deduplicate workers and suppliers across systems, and serve the same number everywhere.
Six teams ask the same question: "How many contingent workers do we have right now?"
metric:
name: active_worker_count
label: Active contingent workers
owner: workforce-program-office
status: certified
version: 2.1
type: count_distinct
entity: worker_master_id # deduplicated across VMS
filter: assignment_status = 'active'
as_of: business_date
sources: [fieldglass, beeline, vndly]
dimensions: [category, job_family, supplier_parent,
region, cost_center]
policies:
pii: aggregate_only # no worker names in BIThe metrics workforce programs argue about most
Each one gets an owner, a certified definition, version history and tests. When the program office changes how "active" is defined, every dashboard, extract and AI answer changes with it the same day.
One supplier, five names. Which one do you pay?
The same supplier shows up differently in every VMS and in accounts payable, so spend, performance and risk get split across "different" companies. Click through the four steps I lead today with AI-assisted supplier master data. Supplier names here are fictional.
Build the capability, not another dashboard
Start with the decisions the program office, procurement and finance need to make. Build reusable data products around them. Use AI where it makes governed evidence easier to explore, and measure whether each intervention created value.
Work with the VMS, warehouse and BI tools already in place; add technology only where there's a real gap.
Supplier and worker identity, and certified metrics, come before any new dashboard or model.
Spend, headcount, supplier and rate products serve many consumers instead of one-off client reports.
Every program has an owner, a baseline and a measured outcome before it grows.
Listen and inventory
- Meet the program office, procurement, finance and HR
- Inventory VMS, AP and HR sources and how each counts workers and spend
- Find where headcount and spend numbers disagree today
Agree the model
- Certified definitions for headcount, spend, rate and time to fill
- Supplier master with parent hierarchy across all sources
- Pick one high-value question, such as rate variance by job family
Prove one thin slice
- Signal → decomposition → supplier or job family → owner
- Plain-language investigator on certified metrics only
- Baseline set so the next supplier program can be measured