AI Evaluation Engineer (Python, QA or Security)
Mindrift • Romania, Poland, Czechia, Hungary, Slovakia, Bulgaria
Education
Not stated
Type
Hourly
Pay Rate
$40/hr
Listed
54d ago
About this role
From the Mindrift listing
Please submit your CV in English and indicate your level of English proficiency. Mindrift connects specialists with project-based AI opportunities for leading tech companies, focused on testing, evaluating, and improving AI systems. Participation is project-based, not permanent employment. What this opportunity involves We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks. You'll create challenging tasks and evaluation criteria within realistic simulated environments: Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history Design tasks from intermediate states of these environments - craft the prompt, define what "solved" means, and ensure the task is solvable by an AI agent Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust What this is NOT Not data labeling Not prompt engineering Not writing code from scratch - the agent writes most of the code; you guide and evaluate What we look for 5+ years in software development Core stack: Python (FastAPI), JavaScript/TypeScript (React), Docker, Postgres, Kafka, Redis Experience writing tests (functional, integration) English proficiency - B2+ Why this is hard Frontier models are already good at coding. Creating a task that genuinely challenges the best models is non-trivial. You need to deeply understand where models fail and what scenarios reveal the difference between a good and a bad solution. Tasks have many valid solutions - writing tests that accept all correct solutions and reject incorrect ones is harder than it sounds. How it works Apply → Pass qualification(s) → Join a project → Complete tasks → Get paid Compensation Up to $40/hr equivalent , depending on level and pace. Tasks are estimated at ~20 hours each; you set your own schedule.
Requirements
- Must be eligible to work in one of: Romania, Poland, Czechia, etc.
- Fluent proficiency in English (Written & Verbal)
- Reliable high-speed internet connection
Interview Prep
This listing calls for this tool directly. Prep for the technical screen:
Why this role
This AI Evaluation Engineer (Python, QA or Security) role pays $40/hr for Software Engineering expertise, and onboarding covers the AI-specific parts of the work you haven't seen before.
Skills and categories
Explore other opportunities in related specializations:
Related jobs
Chatbot Developer (WhatsApp, Telegram, Discord) - Freelance
mindrift • Software Engineering
$60 /hr
Chatbot Developer (WhatsApp, Telegram, Discord) - Freelance
mindrift • Software Engineering
$60 /hr
Chatbot Developer (WhatsApp, Telegram, Discord) - Freelance
mindrift • Software Engineering
$60 /hr
Chatbot Developer (WhatsApp, Telegram, Discord) - Freelance
mindrift • Software Engineering
$60 /hr
Browse All Jobs from Mindrift
Discover more opportunities on Mindrift that match your skills and interests.
View All Mindrift Jobs →Verified Reviews
Community Reviews
Share your experience with Mindrift
Help other candidates make better decisions by leaving a review.
Sign in to leave a reviewLeave your review
Common questions
What does a typical Mindrift task involve?
Read a scenario, write a short prompt for the AI (around 100 words), evaluate two AI responses to it, fact-check the outputs, then write a brief explanation (around 50 words) on which response is better against the project rubric. The focus is clarity, safety, and rule-following, not creative writing.
Who is Mindrift good for?
Freelance writers, editors, and generalist AI tutors with strong English fluency and solid research skills, no specialized tech degree required. Built by data-labeling company Toloka, Mindrift also opens higher-paying specialized projects (cybersecurity, medicine, law) to domain experts once verified.
What does task-based AI training work look like?
Practical, hands-on data work: recording short videos, categorizing images, rating text responses, or analyzing data. Tasks are designed to be short and distinct, typically 5 to 60 minutes each.
What does asynchronous AI training work mean in practice?
No set hours, no check-ins, no meetings. You log in when you want, pick up an available task, complete it, and submit; nobody is waiting on you in real time. That's different from remote employment, where you're expected online during business hours. The tradeoff: you're competing with others for available tasks, so an empty queue means there's simply nothing to do until more work is released.
What does Software Engineering work look like for an AI Evaluation Engineer (Python, QA or Security)?
Tasks here are scoped to Software Engineering, not generic labeling. As an AI Evaluation Engineer (Python, QA or Security), expect to draw on real domain judgment (evaluating outputs, correcting errors, or providing expert reasoning specific to Software Engineering) rather than following a one-size-fits-all rubric. If you don't have hands-on Software Engineering background, this is likely not the right listing to start with.
What specific skills does this listing call for?
AI Training is named directly in the listing. If you don't have hands-on experience with this, expect the screening process to test for it directly rather than accepting adjacent experience as a substitute.
How much does this specific role pay?
This listing is posted at $40/hr, an hourly rate. Pay can change between when we last checked the listing and when you apply, so confirm the current number on the platform's own application page before committing time.
Can I apply from outside Romania, Poland, Czechia and 3 other countries?
This specific role is open only to people based in Romania, Poland, Czechia, Hungary, Slovakia, and Bulgaria. If you are somewhere else, applying is unlikely to lead to an offer even if you pass the assessment, because the restriction is usually about where the work can legally be contracted rather than your skills. Read the full description for any tax-residency or right-to-work caveats before you apply, since they can differ by country.