Founding Senior
Backend Engineer
NEW YORK CITY · IN-PERSON
Casey fills shifts today. Tomorrow, Casey protects authorized hours, maximizes utilization, and owns the revenue cycle from referral to reimbursement. The bridge between “shift filler” and “revenue engine” is the data platform you're going to build.
We operate a sophisticated multi-agent system against four major EMRs (WellSky, HHAeXchange, AlayaCare, Axxess) — 29,000+ providers of serviceable TAM, all of whom chose us as their system of action. Our agents already generate 25–30% of daily activity in some of those platforms. The volume is accelerating.
What we need now is the horizontal data layer that makes every agent decision measurable against customer-level business outcomes — per-client, per-caregiver, per-agency, over weeks and months. That's the foundation our experimentation culture, our ML roadmap, and our RCM ambitions all rest on. You own it end-to-end, and you build the team around it.
YOU OWN
- –The outcome data layer — the instrumentation connecting every agent decision to customer-level business metrics: client retention, caregiver satisfaction, authorization utilization, hours erosion. The foundation for everything our AI/ML work depends on, every ROI claim we make to customers, and every experiment we run.
- –The real-time event backbone — visit status changes, clock-ins, call-outs, ranking decisions flowing as events any system can consume in real-time. This is what makes the EMR polling and rate-limiting problems structurally go away instead of being band-aided.
- –The integration platform — canonical data models across WellSky, HHAeXchange, AlayaCare, and Axxess so that adding EMR #5 is configuration, not 30+ bespoke commits. Also the foundation for RCM: you need unified authorization and billing models before you can build agents that optimize revenue.
- –Analytics and experimentation tooling — warehouse architecture that makes it trivially easy for any engineer to answer "did this change improve customer outcomes?" without writing a custom query every time.
YOU ARE
- –Humble about and genuinely excited by the unglamorous work — data pipelines, ETL, cleaning up messy EMR data, building the joins nobody has written. You see the invisible infrastructure the way a great plumber sees plumbing.
- –Opinionated about data modeling as a craft. FHIR standards, canonical visit/authorization/billing models — not just "I'll build whatever schema you need" but "here's how these entities should relate, and here's why this model compounds downstream."
- –AI-native and technologically curious. A critical field only exists as free text in an EMR web portal? You think: could we use an LLM to parse it into structured data with validation? No API? You think about browser agents. You spot opportunities to apply new capabilities to data infrastructure problems.
THE ARC
THE PROCESS
Intro Call
45 min, remoteDaniel walks you through the role and the data platform gap. You walk us through a data or integration platform you've built and the design decisions you'd make differently now.
Technical Exercise
Async, own timeWe hand you a reconciliation dataset with real-world inconsistency patterns. You design a detection and classification system and propose resolution strategies. Record a Loom walking through your thinking.
On-site Day
Half day, in-officeReview + live extension of your exercise with Ethan and Greg, architecture conversation with Daniel, talent-magnet chat with Alice, close with Victor.
Offer
Target: offer signed by May 10.
APPLY
Send a short note to daniel+backend@zingage.com — walk us through a data or integration platform you've built, the model decisions you'd make differently now, and where you think our data layer should be in 12 months.