Software Engineer

About the role

The product

• Internal practice-management and clinical-operations platform for a telehealth provider. It sits as a middleware layer between our staff and the Healthie EHR
• Python/FastAPI backend + Next.js/React frontend in a monorepo, handling live PHI for real patients - not a greenfield project and not a prototype.
• Small team, high ownership. You will take features end-to-end: data model →API → UI → tests → deploy. You will not be handed pre-decomposed tickets.

What you'll work on

• Patient intake, chart notes, scheduling, and billing workflows used daily by clinical staff.
• Third-party integrations: EHR, clearinghouse/eligibility vendors, CRM, and form webhooks.
• An in-app clinical AI Copilot built on streaming LLM APIs.

Required experience

• Python 3.12 + FastAPI
• TypeScript + Next.js 15 (App Router) + React 19
• Zustand for client state, or other strong global state management experience
• GraphQL as a consumer - writing and debugging queries and mutations against a third-party API you do not control
• REST/HTTP integration
• Unit testing with Vitest and Pytest
• Redis caching
• PostgreSQL / Supabase
• Git fluency

Strong plus

• US healthcare domain knowledge: insurance eligibility (EDI 270/271), payer and plan-type classification, Medicare vs. Medicare Advantage vs. Medicaid, MBIs, CPT/ICD-10, CMS-1500 claims.

LLM application engineering: OpenAI/Anthropic APIs, streaming over SSE,tool/function calling, prompt design, token and cost control, multi-turn session persistence.

• CRM and webhook integration work: Zoho CRM REST v8, OAuth 2.0 client credentials, form webhooks.

Observability: Sentry, PostHog, and log-driven debugging in a hosted environment.

How we work

• Strong written English. Most communication is asynchronous: PR descriptions, design proposals, written status. You need to explain a trade-off in writing without a call.
• Escalates early. When a spec is ambiguous or an approach is blocked, you raise it within hours instead of shipping a guess three days later.
• Reads before writing. The repo carries extensive internal documentation —architecture notes, per-domain summaries, prior implementation plans. Search that context before proposing a design.
• Comfortable using AI coding agents (Claude Code)

as part of the normal workflow — planning, implementing, and verifying with agent assistance, while staying fully accountable for the code you submit.

• Updates internal documentation as part of finishing a task, not as an afterthought.

Not required

• Mobile, Kubernetes/DevOps, ML training, or data-engineering experience.
• Prior US healthcare experience is a strong plus, but not a hard gate for an otherwise excellent Python/TypeScript integration engineer.