What is yokla?

Questions are AI-generated, but generating them is not considered enough on its own: every specialization pack goes through a four-stage verification line — generation, independent review (by a model from a different provider), a third reviewer who knows nothing about the earlier rounds, and the application of verified findings only. Every numerical or behavioural claim is actually executed and measured with the relevant library installed locally; no number reaches production on a guess.

You paste the job ad you are applying to; AI pulls out the technical topics it asks for and gives you 5, 10 or 15 open-ended questions on them. Questions come from the verified question bank; for a topic the bank does not cover, the AI writes the question itself and labels it “AI-generated”. Each answer you write is scored out of 10 against that question’s verified answer key, and you see what was missing and the correct answer right away. Your multiple-choice scores and ranking are not affected.

No. There are no leaked or copied questions. They are written from scratch against the topic and difficulty profile real interviews test, then verified. The goal is not to memorise one company’s question but to measure whether you actually know the topic.

Software developers preparing for interviews or measuring themselves: there is separate content for junior, mid and senior levels. If you are unsure about your level, the placement mode asks questions from three difficulty bands and tells you where you stand. Backend, frontend, fullstack, mobile, DevOps/cloud, database, security, QA, data engineering, AI and ML are all covered.

There are 50 specializations under the fields: Go, Python, Java, C#/.NET, Node.js, PHP, Rust, React, Vue, Angular, TypeScript, PostgreSQL, MySQL, MongoDB, Kubernetes, AWS, Terraform, Kafka, Spark, Airflow, LLM/RAG and more. When you pick one, about a third of the questions come from that specialization and the rest from the field’s general pool.

Your score is compared against other candidates who took the interview in exactly the same configuration (same field, same seniority, same specialization) — like for like. The comparison only appears once that configuration has enough completed interviews; no misleading percentile is produced from thin data.

Both. Every question, option and explanation is written in Turkish and English, and you can switch languages during an interview. The Turkish content is written in Turkish, not translated.