Agentic Systems Researcher

MissionWe are seeking an Agentic Systems Researcher who wants to apply statistics, econometrics, and/or data science to the software quality discipline.This role involves conquering the combinatorial explosion of agentic AI states, by using advanced probabilistic modeling to intelligently sample, prioritize, and generate the most critical test cases for a given AI agent.RoleOverview: This is a highly quantitative, code-forward role focused on building a statistical test prioritization and generation engine for a targeted agentic V&V (verification and validation) system.Combinatorial Optimization: Design variance reduction, heuristic sampling, and Extreme Value Theory (EVT), or other relevant strategies that can prioritize mission-critical and/or black swan failure modes, instead of inefficiently executing potentially infinite test cases.Structural Mapping: Adapt structural econometric models, discrete choice frameworks, state-space models, or other approaches that can probabilistically map the decision trees of AI agents, identifying high-leverage risk surfaces.Coverage Quantification: Develop defensible indices, capture-recapture, or other techniques that can quantify the surface area of tested versus untested agent logic, and provide ISO 9126 and ISO 25010 software quality KPIs and coverage measurements to our users.QualitiesQuantitative Rigor: You love applying econometrics, causal inference, and advanced predictive modeling to messy, real-world systems to manage risk and ensure reliable behavior.Dimensionality Destroyer: You are obsessed with finding the signal in the noise and intuitively understand how to reduce infinite state spaces into manageable, high-value heuristics.Python Pragmatist: You don't stop at academic papers; you like to translate ideas into code that is easy to connect to real-world tech stacks to bring theory into reality.Bonus PointsA background in econometrics, structural choice modeling, or advanced predictive modeling.Experience adapting statistical models to evaluate AI, ML, LLM, and agentic systems.Experience generating novel IP or published research in our domain.ExperienceExtensive background in quantitative analysis, data science, or statistics, with a history of deploying mathematical models into production-grade software.Understanding of software testing, fuzzing, or LLM orchestration.About UsWe are an early-stage, stealth-mode agentic AI QA startup, backed by Breyer Zou Ventures and successful entrepreneurs, investors, and operators such as Scott Sandell, Sheryl Sandberg, Jerry Yang, and KR Sridhar.