Junior Quality Analyst

Before you apply

• We're direct about how we work. Read this section first.
• This is a traineeship before it is a job. 3 months of training, then employment only if you complete it successfully. The full terms are shown before your application starts, and you have to accept them to apply.
• It is for fresh graduates. You should have completed your education within the last 12 months, or be about to. If you already have testing experience, apply to Quality Analyst instead. Same team, different posting, no terms like these.
• We value people who know what they don't know, and say so. That is the whole point of the first 3 months.
• We move fast. Growth happens outside comfort zones. If this sounds harsh, we're probably not a fit. If it sounds like the work you have been waiting for, keep reading.

The role

Allerin builds production AI. Computer vision on factory lines, agentic systems, analytics clients run their week on. Our case studies carry real numbers, and somebody has to stand behind them. That somebody gathers the evidence.

Quality here is not a phase at the end. It is the discipline of proving, honestly, that a thing does what we claim. You will learn to do that on systems that are actually running, for clients who actually measure.

What you'll learn, in order

• How to write a finding somebody acts on. Steps to reproduce, expected against actual, evidence attached, severity you can defend. Most people never learn this properly and it is the foundation of everything else.
• How to decide what to test. Not a checklist. Risk. What would hurt the client most if it broke, and what is most likely to be broken.
• How to explore. The defects that matter live where nobody wrote a test case, and finding them is a skill you build rather than a process you follow.
• How AI systems fail. This is the part almost nobody will teach you. A vision model that is confident and wrong. An agent that answers differently on the same input twice. A model that got worse only on rare cases after an update. Testing those takes evaluation sets and calibrated judgment.
• How to use the tools without trusting them. Generated test cases, self-healing suites, wide regression. You direct them and you audit what they produce, because fluent output is not correct output.
• How to hold a line. When something is broken two days before a deadline, you say so, with proof. That is learnable and we will teach it.

What you must arrive with

We do not require certifications. We do not require testing experience. We are not filtering on your college.

• You notice things. Software that is slightly wrong bothers you and you want to know why.
• You can write clearly. A finding nobody understood did not happen, so this matters more here than almost anywhere else.
• You are willing to be wrong in public and say so. Every good tester has called something broken that was not.
• You can read a little code, or you are willing to learn. You do not need to write it yet.
• You finished your education within roughly the last 12 months, or you are about to.

A GitHub account is required to apply. We are not counting stars. Anything that shows how you work is fine, including a college project or a small script. If your best evidence is a bug report rather than code, the first question below is exactly where to put it.

What we offer

• Systematic evaluation of AI systems, practised on production computer vision and agentic products. There is almost no entry path to this skill anywhere else
• Your findings read and acted on from the first month
• Engineers who take evidence seriously and will tell you when yours is thin
• A clear path. Complete the training, get the job

How we hire

Read the terms, accept them, then apply. Six questions, and the first is a work sample you can do today.

If we talk, we will go through your bug report with you and push on it. That is most of the conversation.

A human reads every application, and answers that read as model-generated are tested in depth in that conversation, where they do not survive.