Senior Software Engineer - AI Agent Evaluation

<img src="https://we-work-remotely.imgix.net/logos/0171/5754/logo.gif?ixlib=rails-4.0.0&w=50&h=50&dpr=2&fit=fill&auto=compress" /> <p> <strong>Headquarters:</strong> Canada <br /><strong>URL:</strong> <a href="https://mindrift.ai/">https://mindrift.ai/</a> </p> <p><em>Please submit your CV in English and indicate your level of English proficiency.</em></p> <p></p> <p>Mindrift connects specialists with project-based AI opportunities for leading tech companies, focused on testing, evaluating, and improving AI systems.&nbsp;<strong>Participation is project-based, not permanent employment.</strong></p> <p></p> <p><strong>What this opportunity involves</strong>&nbsp;</p> <p>We're building a dataset to evaluate AI coding agents - how well a model handles real-world developer tasks.</p> <p></p> <p>You'll create challenging tasks and evaluation criteria within realistic simulated environments:</p> <ul> <li>Build realistic developer environments - a virtual company with codebase, infrastructure, and context (tickets, docs, conversations) that forms a believable development history</li> <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> <li>Write tests that verify agent solutions - accept all valid approaches and reject incorrect ones, neither too strict nor too lenient</li> <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> </ul> <p></p> <p><strong>What this is NOT</strong></p> <ul> <li>Not data labeling</li> <li>Not prompt engineering</li> <li>Not writing code from scratch - the agent writes most of the code; you guide and evaluate</li> </ul> <p></p> <p><strong>What we look for</strong></p> <ul> <li>5+ years in software development</li> <li>Core stack: Python (FastAPI), JavaScript/TypeScript (React), Docker, Postgres, Kafka, Redis</li> <li>Experience writing tests (functional, integration)</li> <li>English proficiency - B2+</li> </ul> <p></p> <p><strong>Why this is hard&nbsp;</strong></p> <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> <p></p> <p><strong>How it works</strong></p> <p>Apply → Pass qualification(s) → Join a project → Complete tasks → Get paid</p> <p></p> <p><strong>Effort estimate</strong></p> <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> <p></p> <p><strong>Compensation</strong></p> <p>Up to <strong>$50/hr equivalent</strong>, depending on level and pace. Tasks are estimated at ~20 hours each; you set your own schedule.</p> <p><strong>To apply:</strong> <a href="https://weworkremotely.com/remote-jobs/mindrift-senior-software-engineer-ai-agent-evaluation">https://weworkremotely.com/remote-jobs/mindrift-senior-software-engineer-ai-agent-evaluation</a></p>