AI Evaluation & Reliability Engineer (Agents & LLM Systems)

abra R&D is looking for a AI Evaluation & Reliability Engineer (Agents & LLM Systems)! abra R&D is looking for a AI Evaluation & Reliability Engineer who will take part in building the next-generation agentic analytics platform, the first real-time database optimized for AI agents at scale. We’re looking for a Senior AI Evaluation & Reliability Engineer to define and build how AI agents are measured, validated, monitored, and improved in production. This role sits at the intersection of LLM systems, evaluation research, and production-grade engineering. You will design evaluation methodologies, build LLM-as-a-judge systems, and develop agent-based testing frameworks to ensure correctness, robustness, and reliability of complex multi-agent workflows operating on real-time data. What You’ll Do: Design and implement evaluation frameworks for AI agents and multi-agent systems Build LLM-as-a-judge pipelines to assess correctness, reasoning quality, and output quality Develop agent-based evaluation systems (agents evaluating agents) for scalable testing Define metrics, benchmarks, scorecards, and methodologies for agent reliability and performance Build data-driven evaluation pipelines using synthetic and real-world datasets Identify and analyze failure modes, edge cases, and non-deterministic behaviors Improve agent robustness, consistency, and reliability in production environments Work with tools such as Google ADK, Opik, and related evaluation frameworks Collaborate closely with AI, platform, and database teams to shape agent–data interaction quality Requirements Must have: 4–8+ years of experience in software engineering, AI systems, or evaluation/QA engineering Strong programming skills in Python Hands-on experience working with LLMs in production environments Experience building evaluation systems, automation frameworks, or testing infrastructure Strong understanding of prompt engineering, tool use, and agent behavior Ability to think in terms of metrics, correctness, and system reliability Nice to have: Experience with LLM evaluation frameworks (Opik, LangSmith, etc.) Experience with Google ADK / agent frameworks Experience implementing LLM-as-a-judge or ranking systems Background in data systems, analytics, or real-time pipelines Experience with multi-agent systems Familiarity with statistical evaluation methods or experimentation (A/B testing, scoring systems)