Data Scientist, Agent Evaluations & Quality

About the Role

This role sits at the intersection of applied data science and AI product quality for a small, fast-moving AI productivity startup building autonomous agents that handle email, calendar, browser, and business software tasks. You will own the measurement of agent quality end-to-end: turning ambiguous product behavior into rigorous, actionable evaluation systems that directly guide engineering and product decisions.

What You'll Do

• Architect and maintain automated evaluation pipelines that measure agent quality across capabilities and product surfaces.

• Translate agent capabilities into explicit success criteria, including pass, partial-pass, and failure definitions for complex multi-step tasks.

• Build representative gold datasets and regression suites covering common workflows, edge cases, ambiguous requests, and adversarial scenarios.

• Define and track metrics such as task success, tool-selection accuracy, instruction adherence, factual consistency, latency, cost, and reliability.

• Design deterministic and model-based graders, calibrate LLM-as-a-judge systems, and measure grader agreement, false positives, and false negatives.

• Analyze traces, tool calls, model outputs, and production outcomes to identify root causes and build a useful failure taxonomy.

• Compare models, prompts, tools, and capability implementations using rigorous offline experiments and production evidence.

• Build dashboards and release-quality signals that make evaluation results understandable and actionable for engineering, product, and leadership.

• Partner with capability engineers to recommend improvements and verify that fixes raise quality without unacceptable regressions in cost, latency, or reliability.

What We're Looking For

• 5+ years in data science, machine learning, or analytics roles, with a focus on evaluation systems, metrics frameworks, or quality measurement for production systems.

• Demonstrated experience designing and implementing evaluation frameworks, grading systems, and success criteria for ML or AI systems in production.

• Strong Python and SQL proficiency with the ability to build automated data pipelines and production-quality analysis code at scale.

• Solid statistical and experimental design knowledge: sampling, variance, uncertainty quantification, bias detection, confounding variables, and significance testing for non-deterministic systems.

• Experience with ground-truth data development: labeling guideline design, annotation quality control, ambiguity resolution, and dataset maintenance.

• Working knowledge of LLM behavior, tool use, retrieval systems, multi-step execution, and practical failure modes of language model systems.

• Ability to connect quantitative patterns to individual system traces and identify failure origins across model, prompt, context, tools, data, and application logic.

• Experience communicating evaluation results, methodology, uncertainty, and trade-offs to both technical and non-technical stakeholders.

• Comfort operating with high ownership in ambiguous, fast-moving environments, independently turning open-ended quality questions into evaluation systems.

• Experience with LLM-as-a-judge systems, agentic or multi-step task evaluation, or benchmarking platforms for AI systems is a strong plus.

Location

On-site in Palo Alto, California, United States. Visa sponsorship is not available for this role.

Originally posted on Himalayas