Software Engineer

$50–100/hr

rate as listed

Workload: ~15 hrs/week
Engagement: Project-based
Location: Worldwide
Field: STEM and Coding

Build reproducible reinforcement-learning environments that test whether AI models can fix bugs, add features, refactor, and optimize real code. You write the task, the containerized repo it runs in, the test harness, and the golden reference solution — in Python, Java, Rust, Go, C++, or TypeScript.

What the work actually is

This is not code review of model output. You are authoring the exam. Each task starts with a realistic software engineering problem — a subtle concurrency bug, a slow query path, a feature that has to slot into an existing architecture, a legacy module that needs restructuring — and you build the whole apparatus around it: a repository state that reproduces the problem deterministically, a Dockerfile or equivalent that builds from scratch every time, a test suite that fails before the fix and passes after, and a golden reference solution with written reasoning explaining why that approach is correct and what the plausible wrong turns are.

The hard part is usually reproducibility and grading, not the engineering. Tests that pass because of a cached artifact, network calls to a package registry that changes next week, a test suite a model can satisfy by special-casing the assertion — these are the failure modes that get environments rejected. You will also review peer-submitted environments and say plainly when a task is under-specified or the reference solution is one of several defensible answers.

What the screen looks for

micro1's process is AI-led and moves fast — they state roles typically fill within 48 hours and expect first tasks within 24–48 hours of onboarding. Expect an interview that asks you to name a specific language and go deep in it: tooling, build system, testing framework, how you'd pin dependencies, how you debugged something genuinely hard. Generic seniority language does poorly; specific war stories with mechanisms and outcomes do well. Prior AI experience is explicitly not required, but demonstrable depth in at least one of Python3, Java, Rust, Go, C++, or TypeScript is.

Logistics

Contractor, remote, fully async, around 15 hours a week
Output-based pay — per accepted task, not per hour; observed rates for this category run roughly $50–100/hr equivalent, not guaranteed
Minimum weekly task submissions apply; ask what the number is before accepting
Time per task varies widely by your workflow, and rework on rejected environments is on you

What they look for

Deep hands-on expertise in at least one of Python3, Java, Rust, Go, C++, or TypeScript
Debugging complex defects and performance bottlenecks in unfamiliar codebases
Building reproducible, containerized build and test environments
Test design — writing suites that verify behavior rather than reward pattern-matching
Refactoring large or legacy codebases without changing observable behavior
Clear technical writing to document reasoning behind a reference solution

Before you apply

Would you pass this screen?

Drawn from what this listing actually requires. Answered in your browser — nothing is sent to us, and we do not store your answers.

Do you have significant professional hands-on experience in at least one of Python3, Java, Rust, Go, C++, or TypeScript?
Core language depth · required
Can you commit approximately 15 hours per week to this contract?
Weekly availability · required
Can you begin your first tasks within 24–48 hours of completing onboarding?
Immediate start · required
Do you accept compensation paid per accepted task, subject to a weekly minimum submission requirement, rather than hourly?
Output-based pay · required
Have you worked on refactoring or modernizing a large existing or legacy codebase?
Legacy refactoring · preferred
Have you delivered a significant feature end to end, from design through production release, on a cross-functional team?
Feature delivery record · preferred

0 of 6 answered

Included with this listing

Interview prep — 10 questions

Matched to what this platform's screening actually measures. Answer each out loud in 60–90 seconds: claim → concrete example → implication.

What it tests: Whether your claimed depth is anchored to real systems in one named language rather than surface familiarity across six.
A strong answer: Name one language, then a system you built or maintained in it with scale, duration, and your specific role — plus the tooling you live in daily (build system, test runner, profiler, package manager).
Where people lose it: Listing four languages to seem broadly useful. The screener follows up in whichever you name first, and thin answers there cost more than a narrower claim would have.

Summary, eligibility guidance and screening preparation are prepared by Sidequest from the platform's posting. Check the source listing before applying.

STEM AND CODING on Sidequest

What stem and coding AI work pays

Rates by specialty, the platforms hiring, and every open role in this field.