Senior Software Engineer, Payroll Engine

de Anda Capital; United States; Remote; USD 100,000 - 105,000; Full time

Job summary

de Anda Capital seeks a Senior Software Engineer to own the payroll engine, domain model, and append-only ledger. You will craft the pay-run lifecycle from draft through commit, ensuring cent-level accuracy across real pay runs.

You’ll implement gross-to-net orchestration, integrate time-and-attendance data, and maintain a regression harness. Remote (U.S.) with a strong emphasis on testable, auditable code and robust data integrity.

Qualifications

• 5+ years building transactional backends in Python with Django and PostgreSQL.
• Experience delivering production-grade systems with ledger-like correctness.
• Fluent in integer-cent arithmetic, rounding rules, and corrections vs reversals.
• Knowledge of state machines and event-sourced/append-only data models.
• PostgreSQL migrations on live data with constraints and transactions.
• Tests as part of change; replay/golden-file regression suites.

Responsibilities

• Own payroll domain model, ledger, and pay-run engine from draft to commit.
• Build an append-only ledger serving as source for registers and GL.
• Develop pay-run state machine: draft, preview, calculate, approve, commit.
• Implement gross-to-net orchestration with tax adapters and deductions.
• Bridge time-and-attendance into earnings lines and job-cost data.
• Maintain regression harness to replay corpus on changes and verify variances.
• Write and review PRD-driven technical docs used by coding agents.

Skills

Python

Django

Integer arithmetic

Code review/PRD adherence

GraphQL

React

Apollo

CI/CD practices

Testing

Tools

PostgreSQL

GraphQL (Graphene)

Celery

Docker Compose

GitHub Actions

Docker

Job description

Senior Software Engineer, Payroll Engine

Team: Payroll

Level: Senior (IC)

Location: Remote (U.S.)

Start: early November 2026

Compensation: $100,000–$105,000

THE ROLE

You own the heart of the product: the payroll domain model, the append-only ledger, and the pay-run engine that takes a run from draft to committed check. Your code decides what people get paid. The exit criterion for your first four months is that the engine reproduces a corpus of real pay runs from the incumbent system to the cent.

WHAT YOU WILL DO

• Design and build the domain model: pay groups and calendars, earning and deduction codes with limits and ordering, employee tax profiles, year-to-date and quarter-to-date accumulators per EIN.
• Build the ledger as an append-only record of every committed check and adjustment, the single source for registers, general ledger, year-end and audit.
• Build the pay-run state machine: draft, preview, calculate, approve, commit; regular and off-cycle runs; manual checks; voids and reversals; retroactive pay.
• Implement gross-to-net orchestration around the tax adapter: pre-tax and post-tax deductions, imputed income, supplemental wages, garnishment calculation with withholding limits.
• Build the generic hours-and-earnings import, then the bridge from the portfolio company's time-and-attendance product that turns locked, approved timecards into earnings lines with rate, pay code, work location and job-cost dimensions.
• Build and maintain the regression harness: replay the corpus on every change, report variances to the cent, and run every variance to ground.
• Write Technical Documents from the product manager's PRDs, direct coding agents against them, and review every agent-authored PR line by line before merge.
• Keep money as integer cents and committed runs immutable; make every calculation re-runnable from stored inputs and engine version.

MUST HAVE

• 5+ years building transactional backend systems in Python; Django and PostgreSQL in production.
• Has shipped a system where money is the output: payroll, billing, banking, insurance claims, accounting. You can explain how you kept a ledger correct under concurrency and corrections.
• Fluent in integer-cents arithmetic, rounding rules, and the difference between a correction and a reversal.
• Designs state machines and event-sourced or append-only data models; knows why committed records are never edited.
• PostgreSQL depth: migrations against live data, constraints, transactions, advisory locks or equivalent.
• Writes tests as part of the change; has built or maintained a replay or golden-file regression suite.
• Has worked with AI coding agents on real deliverables: writes specs an agent can execute and reviews agent output critically.
• Comfortable in a monorepo with CI, Docker Compose and GitHub Actions.

NICE TO HAVE

• U.S. payroll domain: pay codes, pre/post-tax ordering, YTD limits, FLSA overtime, garnishment priority rules, multi-EIN employers.
• GraphQL schema design (Graphene) and React/Apollo for the pay-run UI.
• Celery or equivalent for long-running calculation jobs.
• Time and attendance or scheduling domain; job costing and labor allocation.
• Experience maintaining AGENTS.md-style conventions for a team.

NOT A FIT

• CRUD-app backgrounds with no financial or transactional system experience.
• Engineers who treat floating-point money as fine until it isn't.
• Someone who treats agent output as done once it compiles.
• Someone who wants a green-field stack; this lives inside the portfolio company's existing time-and-attendance codebase.