Integration and data pipelines currently healthcare EDI
Reading this with an agent? llms.txt has the same facts in plain text.
I build the pipelines that move data between systems, and keep them working when things go wrong.
Software development engineer at Etactics in Greater Cleveland. Validation, retries, idempotency and schema change, in healthcare or beyond.
Software development engineer at Etactics, a medical revenue-cycle company in Greater Cleveland. Right now the data is healthcare claims, coverage and payer enrollment. The work is validation, retries, idempotency, ordering and schema change, and it’s the same in any industry. Open to conversations about forward-deployed, solutions and integration engineering roles, in healthcare or beyond.
Event log Replay from the integration failure lab: a recorded run, not live data
Loading the recorded run from the lab…
75%
less logged enrollment-ticket processing time for client managers than the old email workflow, on the self-service payer enrollment intake system my team delivered.
Owner
of the real-time REST API integration behind insurance discovery, and the technical point of contact when product and sales need an answer.
Hackathon win
for a Claude-assisted tool that reproduces production errors, proposes fixes, and generates regression tests.
At work Etactics, Product Development team, since July 2024
On the self-service payer enrollment intake system my team delivered, built customer-facing UI, PDF autofilling and management, and a Jira-webhook Lambda that syncs new issues to the database.
On that system, led setup and design of the in-house Jira API client and the issue-creation AWS Lambda, designed most of the new database tables, and oversaw the migration from the old database.
The intake system has a two-way sync with Jira Service Management, so nobody re-enters data or chases status.
Owned the CSR form-management redesign and the company-aliasing feature.
Write test plans and handle stakeholder communication on healthcare EDI features.
Liaison between Product Development and Operations Engineering.
Contribute to a new claims processing service built on Daffodil (X12/EDI) schemas.
Review peer pull requests, and direct AI coding agents as part of the development workflow.
Stack
Java
Spring Boot
SQL
REST APIs
Python / FastAPI
Docker
AWS
X12 / Daffodil
FHIR / CDS Hooks
The same work, outside healthcare Each thing I’ve built, in the terms other industries use
The healthcare version
The general pattern
X12 EDI claim files, parsed against Daffodil schemas
Batch file ingestion and transformation: parse partner files against a schema, validate them, and hand them downstream.
Real-time insurance-discovery REST API
Synchronous partner API integration: someone else’s contract, auth and response times, with a caller waiting on a latency budget.
Claim acknowledgements that time out, arrive twice or out of order (the failure lab below)
Event pipelines with retries, dead-letter queues, idempotency keys, ordering and schema validation, and the cost of each.
Payer enrollment intake system with a Jira Service Management sync, delivered by my team
Self-service data intake replacing email and CRM workflows, with a two-way sync to the ticketing system so nothing is re-entered.
Payer endpoint discovery (the FHIR CRD Router)
A connection directory for many partner endpoints, with credentials behind a pluggable provider and OAuth2, JWT or mutual TLS per connection.
Payer onboarding workbench
Onboarding diagnostics for a new partner connection: which check failed, and how to fix it.
Financial Planner (personal project)
Bank and card data ingestion: per-source CSV profiles, de-duplication across overlapping imports, daily API sync.
Selected work Personal projects, all open source
A real run on my tower, recorded at phone size: a task in a throwaway scratch folder asks to pip install pytest, I approve it, and all 7 tests pass. Slow stretches are sped up and labelled.
A Python/FastAPI service and mobile web UI for running local and hosted AI agents from one place, with Docker sandboxes, scoped API access, and approval workflows I can answer from my phone.
I built it because the big labs’ agent apps have an uneven mobile experience and remote sessions kept breaking. I wanted one place to track every session, with a fallback when a provider’s usage pool runs low. While I tuned it for local models, I found that a smaller system prompt and a tighter context budget made it faster than popular open-source harnesses.
My own benchmark: 10 hard coding tasks, each run twice, graded mostly by hidden tests. Same model everywhere (Qwen3.6-35B-A3B, 32k context, one RTX 4070 Ti SUPER). The pass counts are within one run of each other, so only the speed gap is a finding.
Hospitals and clinics have to check with each insurer’s system before certain orders. Every insurer runs its own endpoint, so finding the right one is a mess. This Java library maps an insurer’s ID to the right connection, so a provider can keep track of them all in one place. In general terms: a registry of partner endpoints, with credentials behind a pluggable provider.
A self-hosted finance app for a two-person household, built to replace a paid budgeting subscription. It runs on a home server. Its core is a data pipeline: per-bank CSV profiles, de-duplication across overlapping imports, transfer and refund detection, and a daily bank sync, with private accounts kept private.
Keep the directory separate from the patient data. Case study: FHIR CRD Router
Who it’s for: provider organizations that have to reach many insurers’ CRD endpoints. CRD (Coverage Requirements Discovery) is the CDS Hooks workflow where an insurer tells the clinician, at order time, whether something like prior authorization is needed.
The problem
New federal interoperability requirements are pushing payers to stand up CRD endpoints. Each one is another connection to manage on top of the EDI connections we already run: batch file exchanges and real-time ones like eligibility inquiries. Running those connections day to day is labor-intensive, and the legacy process makes it hard to see which connections exist.
The decision
I kept the core to discovery only: it resolves a payer ID to a connection record and never touches patient data, so an organization can adopt it without sending PHI to code outside its own walls. Credentials sit behind a pluggable provider interface, so each organization decides where secrets live and which connections it supports. The alternative I passed on was an all-in-one solution that also collects and maintains the connections.
The tradeoff
It isn’t an all-in-one solution. Organizations still have to collect and maintain their own connection records. The library gives that work a clear structure but doesn’t do it for them.
The companion workbench
A Spring Boot workbench onboards three synthetic mock payers with deliberately different auth and response behavior, and when a connection fails it says which check failed and how to fix it. All three are mocks, so this shows the diagnosis, not real-payer interoperability.
What’s proven so far
Results
Builds and passes 697 tests, plus 16 live tests against the HL7 reference server.
CI runs on Java 17 and 21.
The quickstart runs end to end against an in-process mock CDS Hooks server with synthetic connection records.
Limits
Not yet tried against a real payer sandbox.
The workbench’s payers are mocks.
Next
Test against a real payer sandbox.
Add notifications for credentials that are close to expiring or that fail authentication.
Try it The clinician is waiting: why the 500 ms budget matters
Eligibility checks aren’t enough on their own, because they tell you about the money behind a patient’s coverage, not what’s required for approval. A clinical decision support (CDS) service that answers at the moment of ordering tells the provider exactly what to supply, but a slow one gives that up. Run the demo and watch what happens to the clinician’s screen once the response time passes the 500 ms budget.
Simulated in your browser with synthetic data. Not a real EHR or payer.
MOCK EHR · ORDER ENTRY HOOK order-sign · BUDGET 500 ms
Order
MRI, lumbar spine
Draft
500 ms budget
0100020003000 ms
Clinician’s screen
Sign the order to fire the hook.
Call trace
Ready when you are.
A slow answer can be as good as no answer.
CDS Hooks guidance is that services respond on the order of 500 ms. The spec doesn’t say what an EHR must do when they don’t, so the cutoff and the “sign without guidance” behavior here are illustrative; real EHRs vary. Card actions are inert.
Try it Watch a payer connection break, and see why
Routing a request to the right partner is half the job. The other half is working out why a new connection fails. These are recorded runs of the payer onboarding workbench against two of its three synthetic payers, both healthy and with faults injected on purpose. Pick a scenario to replay its steps, then see the failing check and the suggested fix.
Recorded run against a synthetic mock payer. Passing here does not prove real-payer interoperability.
Integration failure lab Break a pipeline on purpose, then turn on the safeguards
Most of my work is moving data between systems that don’t fail politely. Messages time out, arrive twice, show up out of order, or change shape without warning. The pipeline here carries healthcare claims, but none of those failures is healthcare-specific. Break it on purpose, then turn on the safeguards every integration relies on (retries, dead-letter queues, idempotency, ordering, schema validation) and see what each one costs.
Simulated in your browser with synthetic data. In guided mode you can also replay real runs recorded from a Java 21 / Spring Boot version that executes the same scenarios over HTTP. Source and tests ↗Full case study →
Hi, I’m Daniel.
I studied computer science and engineering with an AI specialization at Ohio State (class of 2024) and joined Etactics that July as a software development engineer on the EDI product team. Since then I’ve owned features such as the CSR form-management redesign and company aliasing, helped my team deliver a payer enrollment intake system, and become the person product and sales bring their integration questions to.
Currently
Iterating on my subscription-agnostic Agent Harness, practicing for disc golf tournaments, and getting ready for hockey season.
I live in the Greater Cleveland area. Off the clock it’s mostly music, disc golf, and hockey. Pick one:
Pick a label for the specifics.
Contact Open to new opportunities, collaborations, and conversations