Founding Engineer

About us

Prometheus Health Systems builds Prometheus Codex. It drafts insurance and government medical forms (disability statements, the Disability Tax Credit, provincial benefits) from documents already in a patient's chart. Every drafted field cites its source. Professional judgments are left to the physician, who reviews and signs every form.

We're physician-founded, based in Kelowna, and in pilot with BC clinics. The product is live: chart upload, OCR and LLM extraction, a review screen, sign-off and publishing all run in production on AWS in Canada.

As our first engineer, you'll work directly with the founder, a practising physician who built the platform, and with our clinical collaborators. You'll take ownership of a working system and decide how it grows. You'll also help hire and mentor the engineers and students who join next.

What you'll own

• Building. Ship features end to end, from the clinic's workflow to production. Explain the technical tradeoffs to the founder and to clinicians.
• The core platform. TypeScript/Node.js services, PostgreSQL models and migrations (including row-level tenant isolation), and AWS infrastructure as code. Deploys are repeatable, observable and recoverable.
• The extraction pipeline. Document ingestion, OCR and LLM extraction into validated, typed facts. It handles duplicates, retries, malformed files, negation, dates, units and conflicting sources.
• Traceability and abstention. Every value stays traceable to its source passage. Nothing unsupported is drafted. High-risk or opinion fields are routed to the physician, never filled silently.
• Form onboarding as data. Each new form is configuration and tests, not new engine code. You'll grow the form library province by province.
• Evaluation. Keep the evaluation sets and filled-form samples that decide whether a release is ready. Track field-level errors, clinician edits, latency and cost.
• Security and compliance. Keep access controls, audit trails, secrets handling, retention and SOC 2 evidence part of everyday engineering, not a separate project.
• Leaving it clearer than you found it. Write down technical decisions, help recruit the next engineers, and make the system understandable to the next person who joins.

What success looks like

By day 30: own the platform

• Set up the full local toolchain and test gauntlet, and ship your first changes through our review, CI and security-scan gates.
• Run a production deploy alongside the founder. Then own the deploy runbook.
• Learn the field rules (what may be drafted, what stays with the physician, what's never offered) and how grounding is verified.
• Take ownership of one live workstream end to end, such as the form-fill pipeline or the patient-questionnaire channel.

By day 60: own a release

• Lead a release from plan to deploy. That includes the filled-form samples the founder reviews before any change to how forms fill.
• Onboard one new form end to end as data: mapping, rules, tests and a filled sample. Prove that no engine code was needed.
• Measure extraction quality on our evaluation set and fix the largest error class you find.
• Take over the incident and recovery runbooks, and complete one backup-restore exercise.

By day 90: set direction

• Make one release decision on evidence: evaluation results, clinician edit rates, latency and cost against agreed thresholds.
• Harden one weak area of your choosing, such as observability, tenant-isolation tests or pipeline resilience.
• Present the next quarter's technical plan, and an onboarding guide for the next engineer.
• Help select and onboard our next hire.

What you bring

• Evidence of shipping and operating production applications end to end, including ownership of hard tradeoffs and production failures.
• Strong TypeScript/Node.js and PostgreSQL skills, and experience operating AWS or comparable cloud workloads.
• Practical experience with LLMs or document processing, including evaluation and failure handling. Sound engineering judgment matters more than any one model vendor.
• Clear writing, good security judgment, and the willingness to learn clinical workflows from the people who do them.
• Comfort with early-stage ambiguity, and hands-on work across product, backend, infrastructure and clinic feedback.

Useful experience: Canadian EMR integrations, FHIR/HL7, PDF forms and OCR, health information workflows, SOC 2, or an early engineering role at a startup. A particular degree isn't required if you can show the skills.

Why join

• Broad technical ownership of a live clinical product.
• Direct influence on product direction.
• Meaningful early equity.
• Software that saves clinicians hours, with clinical safety as an explicit design constraint.

How to apply

Email [email protected] with:

• a short introduction;
• an example of a system you built, including a difficult tradeoff or production failure you learned from;
• a link to code, if you have it.

Selection includes a paid practical exercise on synthetic documents. Please tell us if you need accommodations during the hiring process.