SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Nava is a fast-growing benefits brokerage fusing deep healthcare expertise with cutting-edge technology to modernize the employer-sponsored health insurance experience. The company has built HQ, an AI-powered benefits platform helping HR teams manage renewals, simplify strategy, and empower employees.
You will own the external data pipelines feeding Nava's domain models. Health and benefits data arrives from many sources in many formats: eligibility files from HR and payroll systems, carrier feeds, member IDs, enrollment elections, and claims. Your responsibility is to ingest this data, normalize it to canonical models, then map and validate it so it associates correctly with employers and their people. You are part of a concentrated data team focused on data hygiene, quality, and leverageability across the Nava ecosystem.
Clean, connected data unlocks Nava's AI features, member experience, and data-driven conversations between employers and brokers. It enables employees to understand their benefits and choose the right plan, puts member IDs in hands at point of care, and drives audits that catch billing errors—which occur more often than expected.
In your first year, you will:
- Own ingestion end-to-end: build pipelines that fail loudly on bad input, reconcile counts from source to normalized tables, and surface problems before downstream teams or clients encounter them.
- Unify the census platform: consolidate five years of employee eligibility and elections processes under one set of mapping, validation, and testing practices with proven reconciliation.
- Scale the platform roughly 40× current volume: optimize pipeline processes and revisit architectural decisions (OLAP vs. OLTP storage, event streaming vs. nightly jobs) to absorb growth without proportional cost or runtime increases.
- Make mapping and identity explainable: log and review member ID, eligibility, and claims associations so wrong associations are caught internally and never reach members.
- Build observability and automation tooling: ship TypeScript web applications and AI-assisted features like soft name and plan matching.
- Run production software: implement alerts on failure and data-quality regressions, idempotent re-runs, routine backfills, and secure handling of SSNs and health information.
You will have real ownership on a concentrated data team where your decisions on mapping, storage, and orchestration set practice as the company scales. Work visibly unlocks Nava's AI features and member experience, with a direct line to engineers and leaders making product and platform decisions.
The company uses Dagster, dbt, and Python on Postgres and AWS; TypeScript web applications for observability; Airbyte, Parabola, and Salesforce; and Claude Code, Codex, and Cursor for development.
REQUIREMENTS:
- Hands-on ownership of a production data pipeline other teams depended on (building, inheriting, or owning one stage such as ingestion, normalization, mapping, and validation). Experience with messy external sources and downstream users who noticed when data was wrong is essential.
- Python and SQL depth you can exercise without an AI assistant and use to steer one: joins, indexes, query plans, understanding query performance and whether fixes are real.
- An orchestrator in production (Dagster preferred; Airflow or Prefect are comparable) and Postgres or comparable relational database on a cloud.
- Proof habits: reconciliation, data-quality gates, tests you would trust at 3 a.m.
- AI-assisted development as a daily workflow: knowing what to delegate and when the agent is wrong.
- Care with sensitive data and clear communication with non-technical stakeholders.
- Not required: Dagster specifically, Spark-scale data, a degree, or a specific years-of-experience number. Hands-on dbt is a strong plus, not a gate.
