AI Testing Specialist LLM Evaluation & Qualit

Testing Specialist - LLM Evaluation & Quality

Design and implement model evaluation frameworks using RAGAS, custom

scorers, and automated accuracy benchmarking to validate agent outputs

against structured specifications and Gherkin acceptance criteria.

Develop adversarial prompt testing suites that probe agent behaviour for

hallucinations, prompt injection vulnerabilities, and edge-case failures across the

Refinement, Decision, and Coding Agents.

Build and maintain regression test suites that continuously verify agent behaviour

consistency across model updates, prompt changes, and skill-file modifications

within the LangGraph-based architecture.

Integrate AI-specific test tooling into CI/CD pipelines using GitHub Actions,

ensuring every agent deployment is gated by automated quality, fairness, and

reliability checks aligned with EU AI Act and DORA requirements.

Define and track quantitative quality metrics, confidence thresholds, retrieval

precision, response latency, and drift detection, providing actionable dashboards

to the engineering team.

Collaborate closely with AI/ML Engineers and DevSecOps teams in Germany

and Hyderabad to establish testing standards, review evaluation results, and

continuously raise the quality bar for production agent releases.

Bachelor's degree in Computer Science, Engineering, or a related field, with

demonstrated experience in testing AI/ML systems or LLM-based applications.

Strong programming skills in Python with hands-on proficiency in pytest, test

automation frameworks, and scripting for evaluation pipelines.

Practical experience with LLM evaluation methodologies and tools such as

RAGAS, custom scoring functions, accuracy benchmarking, and adversarial/red

team testing techniques.

Solid understanding of CI/CD integration using GitHub Actions or comparable

platforms, with the ability to embed AI test gates into automated deployment

workflows.

Familiarity with agentic architectures such as LangGraph or LangChain, including

knowledge of retrieval-augmented generation (RAG) patterns and multi-agent

orchestration concepts. Nice-to-have: Experience with regulatory compliance testing (EU AI Act, DORA),

familiarity with vector database validation and containerization

(Docker/Kubernetes).