Will Jewell › Healthcare & clinical data
Health systems · Health tech · Payers · Provider analytics
Interface and warehouse pipelines for healthcare data teams — built, validated, and watched closely enough that you hear about a problem from me, not from a clinician.
A feed fails, the job reports success, and nobody notices until someone asks why last Tuesday is missing.
A source system changes a field. Everything downstream keeps running and quietly means something different than it did last month.
A migration finished. The row counts were close enough. Nobody proved the numbers on the other side match the numbers on this side.
Nine years at HCA Healthcare — the largest hospital operator in the United States, around 190 hospitals. Three in clinical analytics, then six building the data behind an application about 20,000 clinicians used every day, drawing on four separate record platforms — HL7 v2 off a streaming feed. I also validated a warehouse migration end to end.
I am not an interface analyst and I do not sell an integration engine. I build and prove data pipelines, at the volume a health system actually produces.
Email or Slack, reply within one business day. Written updates, no standing meetings. Calls are available and billed separately.
Scope and price agreed before work starts, after the review. You are never billed by the hour for something I got faster at.
I start with read access. Nothing in your systems changes until you have approved what changes and why.
Fixed price, agreed before anything starts. You are never billed by the hour for something I got faster at, and nothing here needs a procurement process to approve.
Read-only look at the systems involved. Written findings: what is wrong, what it takes to fix, in what order, priced.
Freshness, volume, schema and null checks across the pipelines, with alerts, so breakage surfaces before a person notices it.
Row-level and aggregate parity between old and new, itemised, differences explained rather than waved through.
One source ingested and kept current — incremental, tested, documented, failure modes written down.
Monthly: runs executed, exceptions investigated, fixes applied, one written summary. Reply within one business day.
20 hours a month against your backlog, async, no rollover.
Available when something genuinely needs one. Priced, not refused.
Larger scopes get broken into pieces this size rather than quoted as one number. 50% to start, balance on delivery; monthly lines billed in advance. Discounts come from a smaller scope, never a cheaper version of the same scope.
$750, three to five business days. Read-only access to the systems involved. You get back a written scope: what is actually wrong, what it would take to fix, in what order, priced. If you do not like the scope, you keep the document and we stop.
Ongoing work is a monthly retainer against a named number of hours, with the response time written into the agreement.
A perpetual, irrevocable licence to everything I deliver — code, models, documentation — to use, modify and pass on for any purpose. No expiry, no restrictions, no further fee. If I vanished tomorrow you could hand it to anyone and carry on; the documentation is written on that assumption.
I keep ownership of my own general-purpose libraries, test harnesses and templates — the same way any contractor keeps their own tools. They hold none of your data, schemas, business logic or identifying details, and your licence covers your use of anything built with them.
Full terms: how every engagement runs.