Data Engineer

Prysm Health; Santa Clara (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Prysm Health is seeking a data engineer to design, build, and maintain scalable data pipelines that ingest, process, and store clinical data from multiple sources. You will develop data models and schemas to support fast querying and reporting across healthcare datasets.

You'll integrate data from EHRs, labs, and other healthcare apps, ensure data quality, and contribute to ML- or LLM-backed extraction systems, improving accuracy over time.

Qualifications

• Experience building data pipelines ingesting and processing clinical data.
• Strong SQL and coding in TypeScript or Python.
• Experience with ML or LLM-backed data extraction.
• Ability to treat data issues as incidents.

Responsibilities

• Design, build, and maintain scalable data pipelines.
• Develop data models and schemas for efficient querying.
• Integrate data from EHRs and lab systems with quality controls.
• Migrate facilities off incumbent EHR with mapping and cutover tooling.

Skills

TypeScript

Python

SQL

LLM/ML extraction

Data modeling

Tools

PostgreSQL

Prisma

pgvector

AWS

Terraform

NestJS

Job description

What you'll work on

• Design, build, and maintain scalable data pipelines that ingest, process, and store clinical data from multiple sources.

• Develop and iterate on data models and schemas that support efficient querying and reporting.

• Integrate data from various sources, including EHRs, lab systems, and other healthcare applications, ensuring data quality and consistency.

• Migrate facilities off their incumbent EHR with mapping, reconciliation, and cutover tooling that operators can trust.

You might be a fit if

• You have both built the data infrastructure as well as used it to analyze and interpret data.

• Strong TypeScript, or Python and a willingness to work in TypeScript, plus SQL you are proud of.

• You have shipped LLM- or ML-backed extraction and built systems to measure and improve its accuracy over time.

• You treat a dropped or duplicated record as an incident, not a rounding error.

Bonus points

• Healthcare data standards: HL7 v2, FHIR, X12, C-CDA.

• Experience with CMS submissions (iQIES, PBJ) or a claims clearinghouse.

Our stack

Backend

TypeScript, Node, NestJS

Data

PostgreSQL with row-level security, Prisma, pgvector

Frontend

React, Vite, Radix

Infra

AWS (ECS Fargate, RDS, SNS/SQS, Bedrock), Terraform

Quality

Jest, Vitest, Puppeteer acceptance specs, GitHub Actions merge queue