Quality Analyst, Software and AI Systems

A note on what we're looking for

Read this first. We are direct about how we work.

We build production AI: agentic systems, computer vision, and analytics, for enterprises that measure whether it works. Our case studies carry real numbers, and someone has to stand behind them. That someone gathers the evidence, and this is that role. Quality here is not a phase at the end of a project; it is the discipline of proving, honestly, that a thing does what we claim.

AI has changed this job, in both directions. Tools now generate test cases, heal their own scripts, and run regression suites wider than any manual team could. We expect you to use that leverage, not compete with it. What the tools cannot do is decide what matters, explore what nobody specified, judge severity, and say plainly that something does not work when everyone wants to hear that it does. That judgment is the job.

And there is a second job inside it that almost nobody will teach you. Our products are AI systems, and AI systems fail differently. A test that passes or fails is the easy case. A vision model that is confident and wrong, an agent that behaves differently on the same input twice, a model that got worse only on rare cases after an update. Testing those takes evaluation sets, calibrated judgment, and honest evidence. It is one of the most valuable skills in the industry right now, and you will learn and practice it here on real systems.

We do not require certifications; we test ability directly. We do not count years either; a sharp bug report beats a certificate collection every time. What we do expect on this posting is that you have tested something real, for somebody who cared about the answer. If this sounds demanding, it is. If it sounds like the work you have been waiting for, keep reading.

What you'll own

The verification craft

• Decide what to test and in what order, from risk. What would hurt the client most if it broke, and what is most likely to be broken.
• Explore beyond the specification. The defects that matter live where nobody wrote a test case.
• Judge severity honestly and write findings people act on, with reproduction steps, expected versus actual, and evidence attached. One well-documented critical bug beats ten vague ones.

Testing AI systems

• Build and maintain evaluation sets for our agentic and LLM-backed features, and calibrate the automated judges against your own reading.
• Validate computer vision work the way it fails in the field. False positives and false negatives, rare classes, bad lighting, the last twenty percent. Confidence is not correctness.
• Re-test when models or data change, and catch drift before the client does.

Using the machines

• Run AI-assisted testing tools for breadth. Generated cases, self-healing suites, wide regression. You direct them and audit what they produce.
• Evaluate AI-generated tests with the same skepticism as AI-generated anything. Fluent output is not correct output.
• Work inside the automation infrastructure our test automation engineer builds. That role owns the frameworks and the CI; you own what to test, what the evidence shows, and whether it ships.

Honest evidence

• Your output is the truthful state of the system, in writing, every week. What works, what does not, what is untested.
• When something is broken two days before a deadline, you say so, with proof. That spine is the whole point of the role.

This role is NOT for you if

• You want a script of test cases to execute. That work is automated now.
• You lead with certifications. We will test what you can actually do.
• Your bug reports say "it doesn't work" without steps, evidence, or severity.
• You soften findings because someone senior wants a green tick.
• You think AI output can be trusted because it sounds right.
• You need years on a resume to feel allowed to find a real defect. We still do not count years, we count evidence.

New to testing?

If you finished your education within the last 12 months, apply to Junior Quality Analyst instead. Same team and the same work, but it starts with a paid training programme and its own terms, and the questions are written for somebody who has not done this for a living yet. You will be taken more seriously there than here.

How to apply

• Apply through the steps that follow. The questions are built to surface how you test and how you write, not how well you polish a resume. AI-written answers are obvious, and they are rejected.
• Everything you write, we will dig into if we talk.