OverviewWe are building the evaluation backbone for safe, reliable, and efficient agentic engineering in Microsoft Security. This team will create systems that determine when an AI agent, model, prompt, tool, memory strategy, or orchestration pattern is ready to be used in production security and engineering workflows. The role is ideal for engineers who can operate end to end: understand the workflow, design the benchmark, build the harness, implement validators and graders, run experiments, analyze quality and cost tradeoffs, connect results to production feedback, and help teams make evidence-based release decisions.
Why this role matters: Microsoft Security is moving toward agentic engineering systems for security triage, remediation, repo readiness, and scan-to-verified-closure workflows. Evals are the trust system for that shift. They help decide whether autonomy can safely expand, whether a release should stop, and which configuration achieves the required quality, safety, reliability, latency, and cost bar with the lowest practical human-review burden.
Role mission As a
Principal Security Research Manager on the AI Evaluation Systems team, you will build the common evaluation platform and methodology used by MSec agent programs. You will work across evaluation design, platform implementation, test infrastructure, telemetry, measurement, security workflow understanding, and production learning. Your work will make agentic systems measurable, reproducible, governable, and continuously improving.
Responsibilities- Design and build end-to-end evaluation harnesses for agentic security and engineering workflows, including triage, remediation, repo readiness, escalation, tool use, and scan-to-verified-closure paths.
- Create representative benchmark suites and golden datasets that include normal, edge, adversarial, failure-recovery, regression, and production-derived cases.
- Implement deterministic validators, automated graders, trace analyzers, result stores, comparison views, and workflow adapters that make evaluations repeatable and actionable.
- Measure task success, correctness, safety and policy compliance, failure recovery, latency, tool-call behavior, token usage, total cost per successful outcome, and human-review effort.
- Compare models, prompts, tools, memory strategies, policies, and orchestration patterns under consistent conditions and help teams understand quality-versus-efficiency tradeoffs.
- Integrate evaluations into engineering workflows, CI/CD, release gates, and decision processes so material agent changes are supported by reproducible evidence before production rollout.
- Connect offline evaluation results with production feedback, including accepted and rejected outputs, human overrides, incidents, rollbacks, escaped defects, and customer or service-health signals.
- Detect regressions, benchmark drift, evaluator miscalibration, and cases where eval scores improve while real-world outcomes do not.
- Partner with agent builders, product teams, security engineers, data scientists, program managers, and leadership to turn evaluation results into release recommendations and autonomy-boundary decisions.
- Convert learnings into reusable paved paths: playbooks, templates, onboarding guides, dashboards, scorecards, and reference implementations that can scale across MSec.
What you will build - Versioned benchmark suites for security triage, remediation, repo readiness, tool use, escalation, and end-to-end agentic workflows.
- Evaluation runners and harnesses that can replay tasks, capture traces, evaluate outputs, and compare multiple agent configurations.
- Deterministic validation checks for code, policy, security, provenance, ownership, deployment constraints, and workflow-specific correctness.
- Automated and human-in-the-loop grading pipelines with calibration, sampling, and confidence thresholds.
- Pareto-style scorecards that show tradeoffs across quality, risk, latency, tokens, cost, and human-review burden.
- Telemetry and production feedback loops that continuously expand benchmark coverage and keep offline evaluation anchored to real-world outcomes.
- Release-readiness gates and evidence packages that help leaders and product teams decide whether to scale, stop, or redesign an agent pattern.
QualificationsRequired Qualifications:- Doctorate in Statistics, Mathematics, Computer Science, Computer Security, or related field AND 3+ years experience in software development lifecycle, large-scale computing, threat analysis or modeling, cybersecurity, vulnerability research, and/or anomaly detection
- OR Master's Degree in Statistics, Mathematics, Computer Science, Computer Security, or related field AND 4+ years experience in software development lifecycle, large-scale computing, threat analysis or modeling, cybersecurity, vulnerability research, and/or anomaly detection
- OR Bachelor's Degree in Statistics, Mathematics, Computer Science, Computer Security, or related field AND 6+ years experience in software development lifecycle, large-scale computing, threat analysis or modeling, cybersecurity, vulnerability research, and/or anomaly detection
- OR equivalent experience.
- 1+ year(s) people management experience.
Preferred Qualifications: - Doctorate in Statistics, Mathematics, Computer Science, Computer Security, or related field AND 5+ years experience in software development lifecycle, large-scale computing, threat analysis or modeling, cybersecurity, vulnerability research, and/or anomaly detection
- OR Master's Degree in Statistics, Mathematics, Computer Science, Computer Security, or related field AND 8+ years experience in software development lifecycle, large-scale computing, threat analysis or modeling, cybersecurity, vulnerability research, and/or anomaly detection
- OR Bachelor's Degree in Statistics, Mathematics, Computer Science, Computer Security, or related field AND 12+ years experience in software development lifecycle, large-scale computing, threat analysis or modeling, cybersecurity, vulnerability research, and/or anomaly detection
- OR equivalent experience.
- Proven software engineering experience building production systems, developer platforms, test infrastructure, automation frameworks, data pipelines, quality systems, or reliability tooling.
- Ability to design and implement evaluation systems end to end, including task definition, dataset creation, harness implementation, scoring, analysis, and operational integration.
- Demonstrated coding, debugging, system design, and operational excellence skills.
- Experience working with structured data, logs, traces, metrics, APIs, automation workflows, and engineering telemetry.
- Experience with LLMs, AI agents, model evaluation, prompt/tool orchestration, automated grading, evaluation harnesses, or benchmark design.
- Experience with experimentation, statistical confidence, evaluator calibration, regression analysis, human-review protocols, or quality measurement systems.
- Experience with security engineering, vulnerability management, SAST/SCA, SARIF, remediation workflows, secure development lifecycle, or compliance-sensitive systems.
Security Research M5 - The typical base pay range for this role across the U.S. is USD $142,800 - $274,800 per year. There is a different range applicable to specific work locations, within the San Francisco Bay area and New York City metropolitan area, and the base pay range for this role in those locations is USD $188,000 - $304,200 per year.
Certain roles may be eligible for benefits and other compensation. Find additional benefits and pay information here:
https://careers.microsoft.com/us/en/us-corporate-pay
This position will be open for a minimum of 5 days, with applications accepted on an ongoing basis until the position is filled.
