وظائف مطورى واجهة مستخدم

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n<li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n<li>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<\/li>\n<li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n<li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n<li>Not data labeling<\/li>\n<li>Not prompt engineering<\/li>\n<li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n<li>5+ years in software development<\/li>\n<li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n<li>Experience writing tests (functional, integration)<\/li>\n<li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply ? Pass qualification(s) ? Join a project ? Complete tasks ? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at ~20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n<li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n<li>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<\/li>\n<li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n<li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n<li>Not data labeling<\/li>\n<li>Not prompt engineering<\/li>\n<li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n<li>5+ years in software development<\/li>\n<li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n<li>Experience writing tests (functional, integration)<\/li>\n<li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply ? Pass qualification(s) ? Join a project ? Complete tasks ? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at ~20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n <li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n <li>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<\/li>\n <li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n <li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n <li>Not data labeling<\/li>\n <li>Not prompt engineering<\/li>\n <li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n <li>5+ years in software development<\/li>\n <li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n <li>Experience writing tests (functional, integration)<\/li>\n <li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply? Pass qualification(s)? Join a project? Complete tasks? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at approximately 20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n<li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n<li>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<\/li>\n<li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n<li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n<li>Not data labeling<\/li>\n<li>Not prompt engineering<\/li>\n<li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n<li>5+ years in software development<\/li>\n<li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n<li>Experience writing tests (functional, integration)<\/li>\n<li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply ? Pass qualification(s) ? Join a project ? Complete tasks ? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at ~20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n<li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n<li>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<\/li>\n<li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n<li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n<li>Not data labeling<\/li>\n<li>Not prompt engineering<\/li>\n<li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n<li>5+ years in software development<\/li>\n<li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n<li>Experience writing tests (functional, integration)<\/li>\n<li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply ? Pass qualification(s) ? Join a project ? Complete tasks ? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at ~20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n<li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n<li>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<\/li>\n<li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n<li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n<li>Not data labeling<\/li>\n<li>Not prompt engineering<\/li>\n<li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n<li>5+ years in software development<\/li>\n<li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n<li>Experience writing tests (functional, integration)<\/li>\n<li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply ? Pass qualification(s) ? Join a project ? Complete tasks ? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at approximately 20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n<li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n<li>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<\/li>\n<li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n<li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n<li>Not data labeling<\/li>\n<li>Not prompt engineering<\/li>\n<li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n<li>5+ years in software development<\/li>\n<li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n<li>Experience writing tests (functional, integration)<\/li>\n<li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply ? Pass qualification(s) ? Join a project ? Complete tasks ? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at approximately 20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n <li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n <li>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<\/li>\n <li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n <li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n <li>Not data labeling<\/li>\n <li>Not prompt engineering<\/li>\n <li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n <li>5+ years in software development<\/li>\n <li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n <li>Experience writing tests (functional, integration)<\/li>\n <li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply \u2192 Pass qualification(s) \u2192 Join a project \u2192 Complete tasks \u2192 Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at approximately 20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n <li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n <li>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<\/li>\n <li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n <li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n <li>Not data labeling<\/li>\n <li>Not prompt engineering<\/li>\n <li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n <li>5+ years in software development<\/li>\n <li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n <li>Experience writing tests (functional, integration)<\/li>\n <li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply \u2192 Pass qualification(s) \u2192 Join a project \u2192 Complete tasks \u2192 Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at approximately 20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<h2 class="h5">Job description</h2> <div class="t-break" data-jb-field="description"> <span>Please submit your CV in English and indicate your level of English proficiency.<br> Mindrift connects specialists with project-based AI opportunities for leading tech companies, focused on testing, evaluating, and improving AI systems.<br> Participation is project-based, not permanent employment.<br>What this opportunity involves: We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<br> 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: 8+ 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.<br> Creating a task that genuinely challenges the best models is non-trivial.<br> You need to deeply understand where models fail and what scenarios reveal the difference between a good and a bad solution.<br> Tasks have many valid solutions - writing tests that accept all correct solutions and reject incorrect ones is harder than it sounds.<br> How it works Apply → Pass qualification(s) → Join a project → Complete tasks → Get paidEffort estimate Tasks for this project are estimated to take 30 hours to complete, depending on complexity.<br> This is an estimate and not a schedule requirement; you choose when and how to work.<br> Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<br> Compensation: Up to $150/hr equivalent , depending on level and pace.<br> Tasks are estimated at ~30 hours each; you set your own schedule.<br></span> </div>

<p>Verify and audit daily revenue postings from the PMS and POS systems (rooms, F&amp;B, spa, and other outlets) for accuracy and completeness. Reconcile daily revenue reports against cashier summaries, folios, and actual settlements (cash, credit card, city ledger). Reconcile cash clearance and accounts receivable clearance accounts on a regular basis. Reconcile market segment postings and statistical records against system and production reports. Review rate discrepancies, adjustments, allowances, rebates, voids, and complimentary rooms, ensuring proper approval and supporting documentation. Review complimentary rooms, zero-rated rooms, and house-use rooms, ensuring proper approval is obtained and documented in every case. Review room moves, housekeeping discrepancies, and reservation cancellations and creations for accuracy and completeness. Prepare and distribute the Daily Revenue (Flash) Report to the Director of Finance and hotel leadership. Ensure revenue is recorded in the correct accounting period and posted to the appropriate general ledger accounts. Review and follow up on long-stay rate postings, contract renewals, and expirations. Calculate upselling revenue and commissions for payroll processing purposes. Monitor compliance with internal control policies, SOPs, and audit requirements relating to revenue and cash handling. Review user and system log activity to ensure no compliance leaks or unauthorized transactions. Investigate variances or irregularities and escalate findings to the Director of Finance as needed. Support month-end closing by ensuring all revenue schedules are complete, accurate, and properly documented. Maintain and monitor the accounts receivable (city ledger) sub-ledger, ensuring accurate and timely billing to credit account holders. Prepare and dispatch invoices to corporate clients, travel agents, and other credit accounts, with correct supporting documentation. Follow up proactively on outstanding balances and ensure collections are made within agreed credit terms. Prepare pro-forma invoices for cash-basis accounts at least 24 hours in advance. Review cash-basis accounts regularly and confirm collection of payments. Reconcile the accounts receivable sub-ledger to the general ledger on a monthly basis. Monitor credit limits and flag accounts approaching or exceeding approved terms to the Director of Finance. Prepare aging reports and highlight overdue accounts during monthly credit meetings. Work closely with Sales, Front Office, and other departments to resolve billing disputes and discrepancies promptly. Process new credit account applications in coordination with the Director of Finance. Support month-end closing activities related to accounts receivable, including provisions for doubtful debts.</p><p><strong>Desired Candidate Profile</strong></p><ul><li>Bachelor's degree in Accounting, Finance, or a related field.</li><li>Minimum 3 years of experience in hotel income audit and/or accounts receivable, preferably with an international hospitality brand.</li><li>Solid understanding of hotel accounting systems (e.g., Opera PMS, POS systems) and general ledger principles.</li><li>Strong knowledge of hotel revenue cycles and internal control frameworks.</li><li>High attention to detail and accuracy in financial reporting.</li><li>Strong analytical, reconciliation, and problem-solving skills.</li><li>Proficient in Microsoft Excel and hotel accounting/ERP systems.</li><li>Effective communication skills for liaising across departments, clients, and travel agents.</li><li>Able to work independently, prioritize, and meet deadlines in a fast-paced environment.</li><li>High level of integrity, discretion, and confidentiality with financial information.</li><li>Fluent in English; Arabic is an advantage.</li></ul><p>JOINING OUR TEAM MEANS</p><ul><li>Competitive Compensation : A rewarding package tailored to your experience.</li><li>Career Growth : Opportunities for advancement within our dynamic organisation.</li><li>Inclusive Environment : A vibrant and inclusive workplace that encourages collaboration.</li><li>Shared Purpose: Working with passionate people who live our values every day connecting, innovating, curating, and leading the way together.</li></ul><p>Apply CRAFT YOUR CAREER.</p>

<p>Main Objective of the Position u200b Our mission is to Simplify Life. avertra helps highly regulated, customer-centric industries, led by utilities and energy, deliver exceptional experiences through our MiCustomer platform and digital services. This is a hybrid growth role for someone who is equal parts relationship builder, analyst, and problem solver. You will grow client and partner relationships, represent avertra and MiCustomer in the market, and, just as importantly, dig into customer data to find the insight that drives the next opportunity, for the client and for our own product and growth teams. It is the kind of role that suits a consulting or industrial-engineering mind that also loves being in front of people. We are hiring more than one, and everyone starts broad, across accounts, client relationships, partnerships, and our MiCustomer User Group (MUG), so we can see where each person shines before specializing. You will begin by carrying conversations, extracting insight, and following up independently on live accounts. Within your second year, the path leads to owning accounts and pursuits end to end. Because these relationships are largely with US clients, a flexible schedule with some evening overlap is part of the rhythm. You can work US hours if you prefer, or split your day in a way that keeps you present for the meetings that matter.</p><p><b><br></b></p><p><b>Main Job Responsibilities</b></p><ul><li>Accounts &amp; Growth Planning: Research assigned accounts and their industry context, map stakeholders, and identify whitespace and expansion opportunities. Contribute to and, over time, own account plans that connect client goals to avertra's capabilities.</li><li>Client Relationships &amp; Discovery: Build, maintain, and grow assigned relationships across our community, carrying the conversation and following up independently. Run discovery that goes beyond surface answers, surfacing the goals, constraints, and friction that reveal where we can help.</li><li>Customer Data &amp; Insight: Analyze usage and adoption data to spot patterns, risks, and expansion signals across accounts. Translate what you find into clear, actionable insight, both for the client and as voice-of-customer input to product and growth.</li><li>Pre-Sales &amp; Solutioning: Support pursuits with a product-owner mindset, shaping the right solution to a client's need alongside product and engineering. Articulate the value of the MiCustomer platform (MiPortal, MiAgent, MiComm) in terms the client's decision-makers care about.</li><li>MiCustomer User Group (MUG): Onboard new members, drive engagement and follow-ups, and contribute to planning and content, turning the community into insight and advocacy.</li><li>Events &amp; Process: Represent avertra and MiCustomer at conferences and sessions, on-site and virtual, and own the client follow-ups that come out of them. Design repeatable growth processes, templates, and lightweight tooling that make the whole team faster and sharper, and measure the impact.</li></ul><p><strong>Desired Candidate Profile</strong></p><ul><li>Genuinely Curious: You ask the question behind the question, and you know how to turn curiosity into insight that others walk past.</li><li>Value Articulator: You connect what we do to the outcomes a client actually cares about, in plain, compelling language, without hiding behind feature lists.</li><li>Relationship Builder: You bring warmth and energy that make clients want to keep talking to you. In a client-facing role, presence is not optional.</li><li>Analytical: You are comfortable in usage and adoption data, and you can separate signal from noise to find a real opportunity.</li><li>Process-Minded: You see a better way to run things and you design it, an industrial-engineering instinct applied to how growth actually gets done.</li><li>Product-Owner Mindset: In a pre-sales moment you translate a client's real need into a feasible, valuable solution, partnering naturally with product and engineering.</li><li>Knowledge, Skills and Abilities: Strong process design: you can map how work is done and redesign it for speed and quality. Pre-sales and product-owner mindset: you translate client needs into feasible, valuable solutions. Advanced data analytics: you can analyze customer usage and adoption data and turn it into insight for product and growth. Excellent client engagement, discovery, and relationship-building skills, with an engaging, high-energy presence. Strong spoken and written English for US clients</li><li>Education: Bachelor's degree in business, marketing, MIS, engineering, or a related field.</li><li>Experience: Around 2 to 4 years of experience. Exceptional candidates with more experience are welcome. Any external, client-facing experience is required. Experience in customer success or customer relationships, B2B or product marketing, market research, voice-of-customer, or project management is a strong plus.</li><li>Preferences: Background from a top-tier consulting firm (The big 4). Comfort with AI tools (for example Claude Code) is a plus, not a must.</li></ul>

<p>As a Mid- Senior Full Stack Java Developer at Aspire, you will design, develop, and maintain scalable web applications using Java, Spring Boot, and modern frontend technologies. You will work across the full software development lifecycle, developing robust backend services and user-friendly frontend applications while collaborating closely with QA, DevOps, product, and business teams.</p><h4><b><br></b></h4><h4><b>What You Will Do</b></h4><ul><li>Design, develop, and maintain scalable web applications using Java and Spring Boot.</li><li>Develop and maintain RESTful APIs and microservices.</li><li>Build responsive and user-friendly frontend applications using React.js or Angular.</li><li>Integrate frontend applications with backend APIs and third-party services.</li><li>Design, develop, and optimize SQL databases.</li><li>Write clean, maintainable, and reusable code following best practices and design patterns.</li><li>Perform unit, integration, and API testing.</li><li>Participate in code reviews, technical discussions, and solution design.</li><li>Troubleshoot and resolve application and production issues.</li><li>Work closely with QA, DevOps, product, and business teams.</li><li>Participate in Agile ceremonies and contribute throughout the SDLC.</li><li>Support CI/CD pipelines and application deployments.</li></ul><h4><b><br></b></h4><h4><b>What You Will Need</b></h4><ul><li>4–7+ years of professional software development experience.</li><li>Strong proficiency in Java 8/11/17+.</li><li>Strong experience with Spring Boot, Spring MVC, and Spring Security.</li><li>Hands-on experience with Microservices architecture.</li><li>Strong knowledge of REST APIs and web services.</li><li>Experience with React.js or Angular.</li><li>Strong JavaScript/TypeScript, HTML5, and CSS3 skills.</li><li>Strong SQL knowledge and experience with databases such as PostgreSQL, MySQL, or Oracle.</li><li>Experience with Git and version control.</li><li>Experience with Docker and CI/CD tools.</li><li>Good understanding of software architecture and design patterns.</li><li>Experience with JUnit/Mockito or similar testing frameworks.</li><li>Good understanding of Agile/Scrum methodologies.</li><li>Awareness or knowledge of IT security best practices as defined by ISO/SOC or similar standards.</li></ul><p>Be part of a (Remote is here-to-stay) organization</p><p>Work and learn from great minds</p><p>Explore new opportunities to learn and grow every day by attending technical and nontechnical training</p><p>Get market exposure by working with international tech leaders</p><p>Nursery reimbursement benefit</p><p>Attend virtual and onsite international tech conferences</p><p>Exposure to work in an IT environment that adheres to rigorous security and compliance standards defined by ISO / SOC</p><p><strong>Desired Candidate Profile</strong></p><p>4–7+ years of professional software development experience.</p><p>Strong proficiency in Java 8/11/17+.</p><p>Strong experience with Spring Boot, Spring MVC, and Spring Security.</p><p>Hands-on experience with Microservices architecture.</p><p>Strong knowledge of REST APIs and web services.</p><p>Experience with React.js or Angular.</p><p>Strong JavaScript/TypeScript, HTML5, and CSS3 skills.</p><p>Strong SQL knowledge and experience with databases such as PostgreSQL, MySQL, or Oracle.</p><p>Experience with Git and version control.</p><p>Experience with Docker and CI/CD tools.</p><p>Good understanding of software architecture and design patterns.</p><p>Experience with JUnit/Mockito or similar testing frameworks.</p><p>Good understanding of Agile/Scrum methodologies.</p><p>Awareness or knowledge of IT security best practices as defined by ISO/SOC or similar standards.</p>

<p><strong>Position Summary:</strong> The Microsoft Dynamics 365 Finance &amp; Operations (D365 F&amp;O) Administrator is responsible for managing, maintaining, developing, and optimizing the organization's Microsoft Dynamics 365 Finance &amp; Operations ERP platform. This role works closely with Finance and other business departments to ensure the ERP system supports operational and financial objectives through system enhancements, automation, reporting, integrations, and user support. The successful candidate will have extensive experience with Microsoft Dynamics 365 Finance &amp; Operations, Microsoft Power Platform, Power BI, Azure, and ERP system administration. The role also requires strong analytical, documentation, and stakeholder management skills to deliver scalable business solutions. <strong>Key Responsibilities:</strong> <strong>ERP Administration</strong> Administer and maintain Microsoft Dynamics 365 Finance &amp; Operations. Configure system settings, workflows, security roles, and business processes. Manage application updates, patches, and new feature deployments. Ensure system performance, availability, and data integrity. <strong>Business Analysis &amp; Solution Design</strong> Collaborate with business departments to gather requirements. Translate business needs into functional specifications. Recommend system improvements aligned with business objectives and industry best practices. Evaluate and optimize financial processes, including month-end close and consolidation. <strong>Development &amp; Continuous Improvement</strong> Design and implement process improvements within D365 F&amp;O. Automate manual processes and streamline operational workflows. Identify opportunities for increased ERP utilization and efficiency. Test new Microsoft features and business application functionality. <strong>Reporting &amp; Analytics</strong> Develop and maintain reports using Power BI and D365 reporting tools. Automate recurring reports and dashboards. Support management with accurate operational and financial reporting. <strong>Integration Management</strong> Support integrations between D365 F&amp;O and other enterprise applications. Coordinate with technical teams and external vendors to ensure seamless data exchange. Monitor and troubleshoot integration issues. <strong>Change &amp; Release Management</strong> Manage application changes following established change control procedures. Coordinate testing, deployment, and validation of new configurations. Maintain documentation for changes, configurations, and security updates. <strong>User Support &amp; Training</strong> Respond to D365 helpdesk tickets and provide user support. Troubleshoot system issues and coordinate resolutions. Create user guides, training materials, and technical documentation. Conduct knowledge transfer sessions for business users. Produce and maintain: Business process documentation Functional specifications Technical specifications Solution designs Configuration documentation Security protocols Training materials Report specifications <strong>Internal &amp; External Relationships:</strong> <strong>Internal:</strong> Executive Leadership Finance Department Business Managers IT Team End Users <strong>External:</strong> Microsoft Dynamics 365 Implementation Partner (VAR) Third-party vendors Technology consultants</p><p><strong>Desired Candidate Profile</strong></p><p><strong>Required Qualifications:</strong> Bachelor's degree in information systems, Computer Science, Business Information Technology, or a related discipline. <strong>Experience:</strong> 2 4 years of experience in ERP administration or business systems. Extensive hands-on experience with Microsoft Dynamics 365 Finance &amp; Operations. Experience implementing ERP enhancements and process improvements. Experience supporting enterprise business applications. <strong>Technical Skills:</strong> Microsoft Dynamics 365 Finance &amp; Operations (F&amp;O) Microsoft Power BI Microsoft Power Platform Microsoft Azure SQL/SQL Server Microsoft Excel Microsoft Word Microsoft PowerPoint ERP reporting tools Business process automation Data integration Security and user role administration <strong>Preferred Qualifications:</strong> Knowledge of Generally Accepted Accounting Principles (GAAP). Experience working in Finance or Accounting environments. Experience with ERP implementation projects. Familiarity with software development lifecycle (SDLC) and change management. Experience working with project teams and cross-functional initiatives.</p>

<p><h4>Description</h4> <p>Joining the Business Analytics team to lead requirements engineering across government services and digital transformation programs. The role translates government business needs into structured, build-ready backlogs, models end-to-end processes, and uses AI tools to accelerate analysis and documentation while maintaining quality and data-sensitivity discipline.</p> <h4>Requirements</h4> <h4>Key Responsibilities</h4> <ul> <li>Lead end-to-end requirements elicitation, analysis, and documentation for government eServices and digital transformation initiatives.</li> <li>Produce high-quality BRDs, user stories/PBIs with clear acceptance criteria, BPMN 2.0 process models, and full traceability for Azure DevOps delivery.</li> <li>Facilitate workshops with government stakeholders, resolve conflicting requirements, and obtain formal sign-offs.</li> <li>Map as-is/to-be processes; identify dependencies, edge cases, and non-functional requirements (security, privacy, audit, performance).</li> <li>Apply AI tools (LLMs, prompt engineering, AI-assisted documentation) to speed up analysis and drafting, with sound judgment on validation and sensitive government data.</li> <li>Support UAT, bridge business and technical teams, and ensure delivered solutions meet business objectives.</li> <li>Mentor junior Business Analysts and uphold the team’s BA methodology and standards.</li> </ul> <h4>Required Qualifications & Experience</h4> <ul> <li>Bachelor’s degree in Computer Science, IT, Business, or a related field; CBAP (or equivalent BA certification) strongly preferred.</li> <li>6+ years of hands-on Business Analysis experience, with a strong track record in government/public-sector digital services in the UAE/GCC.</li> <li>Proven greenfield requirements experience — not limited to ERP configuration or pre-sales.</li> </ul> <h4>Required Skills</h4> <ul> <li>Strong requirements elicitation, analysis, and structured documentation (BRD, user stories, acceptance criteria).</li> <li>Process modelling: BPMN 2.0, swimlanes, as-is/to-be (Visio, Lucidchart, Camunda, or draw.io).</li> <li>Backlog management and Agile delivery with Azure DevOps (or Jira).</li> <li>Excellent stakeholder management, facilitation, and bilingual communication (Arabic & English).</li> <li>Hands-on AI proficiency: prompt engineering and AI-assisted analysis/documentation, with disciplined validation and data-sensitivity judgment.</li> <li>Familiarity with eServices, digital-government standards, and the UAE government context.</li> </ul> <h4>Preferred (nice to have)</h4> <ul> <li>Experience with ECM/document management, omni-channel service delivery, and UAE PASS integration.</li> <li>Wireframing/prototyping (Figma) and basic data/BI awareness (Power BI).</li> </ul></p><p></p>

<p><h4>Description<\/h4>\n<p>Please submit your CV in English and indicate your level of English proficiency.<\/p>\n<p>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.<\/p>\n<h4>What this opportunity involves<\/h4>\n<p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<\/p>\n<p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:<\/p>\n<ul>\n <li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history<\/li>\n <li>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<\/li>\n <li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient<\/li>\n <li>Iterate on tasks and tests based on QA feedback - review agent solutions, analyze failures, and refine until the evaluation is fair and robust<\/li>\n<\/ul>\n<h4>What this is not<\/h4>\n<ul>\n <li>Not data labeling<\/li>\n <li>Not prompt engineering<\/li>\n <li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate<\/li>\n<\/ul>\n<h4>What we look for<\/h4>\n<ul>\n <li>5+ years in software development<\/li>\n <li>Core stack: Python (FastAPI), JavaScript\/TypeScript (React), Docker, Postgres, Kafka, Redis<\/li>\n <li>Experience writing tests (functional, integration)<\/li>\n <li>English proficiency - B2+<\/li>\n<\/ul>\n<h4>Why this is hard<\/h4>\n<p>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.<\/p>\n<h4>How it works<\/h4>\n<p>Apply? Pass qualification(s)? Join a project? Complete tasks? Get paid<\/p>\n<h4>Effort estimate<\/h4>\n<p>Tasks for this project are estimated to take 20 hours to complete, depending on complexity. This is an estimate and not a schedule requirement; you choose when and how to work. Tasks must be submitted by the deadline and meet the listed acceptance criteria to be accepted.<\/p>\n<h4>Compensation<\/h4>\n<p>Up to $50\/hr equivalent, depending on level and pace. Tasks are estimated at approximately 20 hours each; you set your own schedule.<\/p><\/p><p><\/p>

<section><p class="heading jdMain">Job Description</p><p class="heading">Roles &amp; Responsibilities</p><div class="paragraph"><br>Support the team with a wide range of finance processes, having hands-on practical daily work and operations as well as being on the front line with internal and external parties such as colleagues in HQ and in the field offices. Support the different teams on a wide range of tasks related to the Accounts Division and analysis of financial data. Support month-end and year-end closing activities, including reconciling accounts. Assist with compilation and consolidation of data. Review supporting documents ensuring completeness and accuracy. Support with system development and process improvement initiatives. Assist in implementing and monitoring internal controls. Assist with development of training materials (in-presence and online) and guidance notes. Assist in drafting and reviewing different types of financial and analytical reports and minutes during meetings. Participate in the preparation of financial reports, presentations, and dashboards for internal and external stakeholders. Collaborate on special projects, providing support as needed, such as in process improvement, data analysis, and policy development. Be responsible for delivering the assigned tasks in a collaborative, timely and efficient manner. Have great opportunities for learning and expanding knowledge and experience within the areas of finance as well as process improvements and system developments. Perform other duties as may be assigned by the supervisor(s).<br>How to apply<br>https://careers.un.org/jobSearchDescription/278872?language=en<br></div></section>

<h2 class="h5">Job description</h2> <div class="t-break" data-jb-field="description"> <span>Please submit your CV in English and indicate your level of English proficiency.<br> Mindrift connects specialists with project-based AI opportunities for leading tech companies, focused on testing, evaluating, and improving AI systems.<br> Participation is project-based, not permanent employment.<br> What this opportunity involves We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<br> 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.<br> Creating a task that genuinely challenges the best models is non-trivial.<br> You need to deeply understand where models fail and what scenarios reveal the difference between a good and a bad solution.<br> Tasks have many valid solutions - writing tests that accept all correct solutions and reject incorrect ones is harder than it sounds.<br> 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.<br> Tasks are estimated at ~20 hours each; you set your own schedule.<br></span> </div>

<h2 class="h5">Job description</h2> <div class="t-break" data-jb-field="description"> <span>Please submit your CV in English and indicate your level of English proficiency.<br> Mindrift connects specialists with project-based AI opportunities for leading tech companies, focused on testing, evaluating, and improving AI systems.<br> Participation is project-based, not permanent employment.<br> What this opportunity involves We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.<br> 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.<br> Creating a task that genuinely challenges the best models is non-trivial.<br> You need to deeply understand where models fail and what scenarios reveal the difference between a good and a bad solution.<br> Tasks have many valid solutions - writing tests that accept all correct solutions and reject incorrect ones is harder than it sounds.<br> 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.<br> Tasks are estimated at ~20 hours each; you set your own schedule.<br></span> </div>