Middle QA Engineer

AI & LLM

DBF is looking for a Middle QA Engineer to join the team for an AI project.

We are developing LLM and RAG-based solutions and are looking for a QA engineer to help ensure the quality and stability of the AI product — from API and backend logic to the quality of the neural network's responses.

💻 About the vacancy

Position: Middle QA Engineer Focus: AI / LLM / RAG Experience: from 3 years in manual testing Format: remote work Employment: full-time Employment type: Contract / Sole Proprietorship Salary: discussed individually Time zone: GMT + 3

🔍 What you will be doing

— conduct functional, regression, smoke, UAT, and negative testing; — test APIs using Postman / Insomnia; — write test cases, checklists, and bug reports in Jira / Zephyr; — work with SQL, logs, and tracing; — test AI chats and multi-turn scenarios; — test edge cases, adversarial / jailbreak prompts, and safety filters; — evaluate the quality of LLM responses: accuracy, relevance, completeness, and tonality; — test RAG pipelines and the correctness of sources in responses; — use AI tools for automating routine QA tasks and writing small scripts.

🧠 What we expect

QA fundamentals: — from 3 years of experience in manual testing; — understanding of basic testing types; — Jira / Zephyr; — SDLC / STLC, Agile / Scrum; — Postman / Insomnia; — basic SQL; — working with logs and tracing.

AI / LLM: — understanding of LLM principles: tokens, temperature, top-p, context window; — understanding of prompt engineering; — knowledge of typical LLM problems: hallucinations, context loss, nondeterminism, sensitivity to phrasing; — experience or understanding of testing multi-turn dialogues, edge cases, and jailbreak prompts; — understanding of AI response quality evaluation; — basic understanding of RAG: chunking, embeddings, vector search, reranking, retrieval; — ability to check the correctness of sources and citations in RAG responses.

Technical skills: — practical experience with Claude Code or similar AI coding CLIs; — basic understanding of OOP; — ability to read autotest code and write simple helper scripts in Python / Java / C#.

🤖 Why this might be interesting

This is an opportunity to work not only with classic QA but also with modern AI products, where quality cannot be checked solely by the principle of "the button works / the button doesn't work."

You will need to understand why the AI gave a specific answer, where it got the information from, how correct it is, and how the system behaves in non-standard scenarios.

📩 @dbf_hrbot