Key responsibilities
Task Execution & Implementation
- Implement clearly scoped tasks within the harness function — including connector maintenance, test case execution, and documentation updates — with minimal errors and on time. Follow instructions carefully, ask questions when unclear, and escalate promptly when something looks wrong.
Agent Monitoring & Issue Flagging
- Monitor agent outputs and behaviour in assigned test scenarios. Log anomalies, failures, and unexpected outputs in a structured way. You are not expected to diagnose root causes independently — your job is to observe accurately, document clearly, and surface issues to the right person without delay.
Documentation & Knowledge Capture
- Maintain clear, accurate documentation of test cases, integration steps, and agent behaviour observations. Help build the team's institutional knowledge as this function scales. Good documentation at this stage has outsized value — treat it as a first-class deliverable.
User & Community Signal Collection
- Support the collection of user feedback and product signals under direction — running structured surveys, collating responses, and organising findings in a usable format. You are gathering and structuring data, not yet drawing strategic conclusions from it.
Learning & Self-Development
- Proactively seek to understand the tools, concepts, and systems you work with — not just how to use them, but why they are built the way they are. Seek regular guidance and feedback. Bring questions, not just status updates. Your growth trajectory is part of what this role is designed to accelerate.
REQUIREMENTS
What we're looking for
- Fresh graduate or up to 1 year of experience in a technical, product, or operations role. We are hiring for potential and attitude, not a lengthy CV.
- Basic coding ability in at least one language (Python preferred). Comfortable with structured tools and workflows — you do not need to be a software engineer, but you must not be afraid of a terminal.
- A basic working understanding of what AI agents are and how they operate — what a prompt is, what a tool call does, what it means for an agent to fail. You don't need to have built one; you need to have used several and paid attention.
- Strong attention to detail and reliable follow-through. You complete what you commit to, flag blockers early, and do not let tasks quietly go stale. These are not soft skills — at this level, they are the job.
- Ability to understand how your work connects to a broader goal — even when you are executing a narrow task, you can articulate why it matters and flag when something seems misaligned.
- Genuine personal curiosity about agentic AI — you use agent tools in your own life, not just as coursework. Side projects, personal experiments, or a demonstrated habit of tinkering with AI tools are strong signals.
- Clear written communication. You will document your own work, write structured observations, and contribute to shared knowledge bases. This is not optional at any seniority — it starts here.
Nice to haves
- Exposure to software quality assurance, testing, or QA workflows — even in an academic or hobbyist context.
- Familiarity with source control (Git) and comfort working in a structured codebase, even without deep engineering experience.
- Experience using tools like Claude Code, Cursor, Manus, or any LLM-based agent product with enough depth to have formed opinions about what works and what doesn't.
- Any background in data analysis, product operations, or user research — even part-time, freelance, or through coursework.
