Staff Data Engineer
Confirmed on the employer's own hiring board on Oct 11, 2026. First seen by Alion on Aug 21, 2026.
About Metabody
Your body runs continuously. Your care runs on appointments - fifteen minutes against half a million minutes of signal a year. The data piles up in apps nobody reads between visits, and by the time anyone looks, the window has closed.
Metabody closes the loop. We unify every signal - wearables, labs, records - into one longitudinal graph; our AI co-clinician reads it continuously and surfaces what matters; and a licensed provider in any of 50 states is ready to act. Not just a wearable. Not just telehealth. The whole loop, compounding with every outcome.
We're early, with real patients across all 50 states and a founding team that's built, shipped, and scaled before. This is a founding-team hire: you'll own a function from day one and set the patterns the next ten people inherit. We hold a high bar - for precision, for follow-through, and for treating patients like people in command, not cases to manage.
The Role
You build the longitudinal graph the entire company - and our AI co-clinician - reads from. This is a doer role: more time in the pipeline than in meetings about the pipeline. You'll stand up our lakehouse with quality enforced where data enters, not patched where it lands - and with PHI handled correctly from the start.
What you'll do
Build the longitudinal health graph end to end - wearables, labs, records, and FHIR-native integrations flowing into one model, with data marts and clean PHI separation that holds up to scrutiny.
Shift quality left: data contracts at ingestion - schema, semantics, ownership, SLAs - enforced where data enters, with bad rows quarantined loudly instead of discovered downstream three weeks late.
Own the ELT and rETL pipelines that serve the business today - and shrink what they have to clean up. Build once, serve many: every transformation you move toward the source is rework the company stops paying for.
Make the numbers governed and agent-ready: metric definitions in version-controlled code with lineage, one semantic interface that analysts and our AI co-clinician both read - policy enforced before a query ever runs. The rest of the company stops second-guessing the numbers.
Direct AI agents on real pipeline work daily - spec the change, review the draft like a junior engineer's PR, and decide what an agent may fix on its own versus what wakes a human. You verify the output and own the result.
What we're looking for
Strong EDW and data lake/lakehouse fundamentals, implemented in production - Databricks ideally.
Healthcare PHI/HIPAA experience: you've built systems where patient data separation was non-negotiable. Experience with EHR (electronic health records) data is a strong plus.
Proven hands-on dbt, ELT, and rETL - shipped and maintained, not read about - plus the judgment to place a transformation at ingestion, in the stream, or in the lakehouse based on consumers and rework, not dogma.
You've moved quality to the source somewhere real: data contracts, producer-side enforcement, schema registries, quarantine paths - and can say what stopped breaking after. Streaming experience is a nice-to-have.
Critical AI fluency: you direct agents on real work, review what they draft with full comprehension, and notice when they're confidently wrong. Semantic-layer or MCP exposure is a nice-to-have.
A builder's temperament. You'd rather ship the pipeline than chair the meeting about it.
Values fit: precision, lead by example, follow-through, high trust and agency, own-it, ruthless empathy.
What we offer
Competitive salary, meaningful early-stage equity, full ownership of a function from day one, and excellent health benefits - traditional insurance plus modern perks.