📝 Django SWE / Terminal-Bench, are you there?

Hey there! We're looking for a Django Software Engineer to join our Intelligence team and build hard backend coding tasks that test the most capable AI models, on a task-based contract. If you are detail-oriented, self-driven, and hands-on with Django and Celery, keep reading!

🌎 100% Remote 💵 Paid in USD per task 📍 LATAM - Reporting to a fully async team 🕒 Task-based contract 🇬🇧 Advanced English

About Athyna 🙋🏻

Athyna is a Global Talent Platform, helping align world-class talent with the best employment opportunities across the globe. We believe in the future of work. That's why we are a 100% globally distributed team across 5 continents. We believe in people, and we value freedom, integrity, quality, and sustainability. We're proudly carbon neutral as an organization. We're very keen on social impact - we truly care about everyone on our team, and we want to make the world a better place for all of us.

What the work is 🧪

Athyna builds hard, original coding tasks used to test frontier AI models. For this role, the tasks live in a real Python web backend stack: Django, Celery, PostgreSQL and Redis. Each task is a small, self-contained problem: a starting environment, an instruction, a reference solution, and hidden tests. A good task is one a skilled engineer can solve in a few hours, but a frontier AI model fails, for a reason you can name.

In your day-to-day, you will 👇🏼

Design and build backend tasks based on real production problems (retries, race conditions, live migrations, stale caches), following our guidelines and templates.
Complete a paid trial task first, so we can see how you work.
Submit each task as a pull request on a private GitHub repo.
Ask and answer task questions in the PR, and iterate on reviewer feedback within the 3-version limit. Many first versions get solved by the model, so expect to make them harder.
Write hidden tests that a plausible but wrong fix would fail.
Write the task instructions yourself, in clear and precise English, without AI help.

Skills you have 🚀

Several years of hands-on backend engineering (professional or research) with real production debugging experience.
Solid experience with Python and Django (DRF is a plus), and PostgreSQL.
Hands-on experience with Celery and Redis in production, including retries, idempotency and task failures.
Comfortable with Linux, Docker and Git.
Able to write a clear, precise problem statement in English.
Curious about where AI models go wrong and patient enough to test it.

Skills that will make you stand out 💥

Experience with SWE-bench, Terminal-bench or similar AI coding benchmarks (a plus, not required).
Experience writing automated tests (pytest) or validation scripts.
Experience with AI evaluation or training data work.
Experience debugging complex environments and live-system migrations.

The process 🧭

A short AI interview, then a shortlist, then one paid trial task (USD 70 once approved), built and reviewed on GitHub. With one approved task, you join the project as a contributor and are paid USD 70 for each approved task.

The screening question ❓

Describe a production bug in a Python web backend (Django, Celery, Postgres or Redis) that took you more than a day to solve. For example a Celery task that retried and charged or wrote twice, a migration on a live table, or a cache that served stale data. What was the cause, and how did you find it? Why would an AI model likely get it wrong, and what would convince it that it was done? What test would catch a wrong fix?

Please pick a bug you are free to talk about. If it is confidential, you can anonymize it by leaving out company names and sensitive details. Answer in your own words, without AI help.