Senior Full-Stack Engineer in United

• 3-5 years shipping production web software, with hands-on experience on both fronts: TypeScript/React and Python/Django. \n

• Designed and consumed GraphQL schemas in production (Graphene or equivalent; Apollo client-side), including codegen and cache behavior. \n

• PostgreSQL fluency: migrations against live data, composite indexes, constraints, transactions, advisory locks or equivalent concurrency controls.

You will own vertical slices of the platform end to end - GraphQL schema, Django services, migrations, Apollo-consuming React UI, and tests on both sides - inside a spec-driven, agent-assisted workflow. The backend is the harder half: calculation correctness, concurrency, multi-tenant isolation, and time handling are where the real work is.

What you will do\n \n

• Turn refined PRDs into Technical Documents precise enough for a coding agent to implement: schema, service boundaries, data changes, acceptance criteria, and the existing repo pattern to follow. \n

• Direct coding agents to produce the implementation, then review their PRs line by line against the Technical Document, AGENTS.md, and acceptance criteria before merge. Catch the tenant-isolation and timezone bugs an agent will not. \n

• Design GraphQL schema the frontend can consume without over-fetching; keep tenant scoping as a first-class argument and index. \n

• Write and harden service code that holds under concurrency: idempotent enqueue to the calc queue, compare-and-swap on state transitions, Decimal arithmetic for hours and money, timezone-correct pay-period boundaries. \n

• Add and extend Celery tasks and Beat schedules (dispatch, delivery, reconciliation) that stay testable without a running worker. \n

• Bring remaining admin screens onto live GraphQL contracts. \n

• Keep CI honest: regression suite green, no coverage erosion. \n

• Deploy to dev via GitHub Actions; work with our DevOps engineer on a limited basis on Terraform and the gated prod deploy. \n

• Triage Sentry across the three apps and the Celery services. \n

Projects you would land in\n \n

1. Period close and payroll delivery - finalization, blocking-vs-warning exception gates, deliveries, post-close adjustments. \n

2. Calc queue completion - per-tenant concurrency caps, scope expansion, config pinning, state CAS, reaper hardening. \n

3. Time-off finalization into payroll - accrual ledger to delivered hours. \n

4. Scheduling - generate and publish from templates, conflicts, swap/claim/drop, schedule-vs-actual matching. \n

5. Admin: locked to live - compensation models, premiums, rates, GL mapping, export mappings, messaging. \n

6. Time capture - kiosk flows, geofence enforcement, device enrollment, punch verification. \n

7. Production readiness - supporting our DevOps engineer on prod Terraform, multi-AZ RDS, secrets rotation, environment-gated deploys. \n, * Built or maintained background job systems (Celery or equivalent): retries, idempotency, dead-lettering, periodic scheduling. \n

• Multi-tenant data isolation and role-based authorization as first-class design concerns. \n

• Correct handling of time: timezones, DST, period boundaries, Decimal hours. \n

• Has worked with AI coding agents on real deliverables: writes specs an agent can execute, and reviews agent output critically rather than rubber-stamping it. \n

• Writes tests as part of the change: Jest + Testing Library and pytest + factory-boy. \n

• Comfortable in Docker Compose, a pnpm/NX monorepo, and GitHub Actions. \n

• Reads unfamiliar code before writing new code; can cite the existing pattern being followed. \n

Nice to have\n \n

• Django Channels / GraphQL subscriptions; WebSocket delivery to React. \n

• Terraform on AWS (ECS Fargate, RDS, ElastiCache, CloudFront); Serverless Framework. \n

• Time and attendance, scheduling, or payroll domain: overtime rules, meal/rest compliance, pay codes, accruals, prevailing wage, certified payroll. \n

• JWT/TOTP authentication, rate limiting behind proxy chains, secrets rotation. \n

• Stripe, SendGrid, WeasyPrint, OpenAI API. \n

• react-intl / formatjs, Storybook-driven UI work, accessibility. \n

• Experience maintaining AGENTS.md-style conventions or agent tooling for a team. \n

Not a fit\n \n

• REST-only backgrounds with no schema-design experience. \n

• “Full-stack” meaning frontend plus a thin API. \n

• Engineers who consider a fixture-backed screen shipped. \n

• Someone who treats agent output as done once it compiles. \n

• Someone who expects infrastructure to be entirely someone else’s problem; DevOps support is limited. \n