Are you interested in operating as a senior scientific leader—owning truth, rigor, and decision quality for complex business problems? Fidelity Institutional’s AI Center of Excellence (AI CoE) is seeking a Principal Data Scientist to serve as a highly tenured individual contributor and domain authority in data science, quantitative modeling, and advanced analytics. This role is intentionally Data Science–first, with emphasis on hypothesis-driven analysis, statistical rigor, causal reasoning, and decision science. The Principal Data Scientist is accountable for what the model means, whether it is correct, and whether it should be trusted—not for building or operating production systems.

Lead hypothesis-driven analyses to answer high-impact strategic and business questions; Design, develop, and evaluate statistical, econometric, and machine learning models where appropriate; Ensure models are theoretically sound, empirically validated, interpretable, and fit-for-purpose; Review and challenge modeling approaches for bias, stability, assumptions, and misuse; Define how success should be measured for complex analytics and AI-enabled initiatives; Design robust evaluation frameworks including offline validation, back-testing, and live measurement; Ensure stakeholders can distinguish correlation from causation in analytical results; Elevate analytics from prediction accuracy to decision quality and business impact; Design and review experiments including A/B tests, quasi-experiments, and observational studies; Apply causal inference techniques (e.g., uplift modeling, DiD, matched controls) to assess incrementality; Guide best practices for power analysis, inference, and result interpretation; Serve as a subject-matter expert on “What worked, why, and by how much?”; Design statistically grounded, interpretable segmentations with clear hypotheses and stability checks; Develop probabilistic and causal models to inform prioritization and intervention strategies; Guide recommendation logic rooted in statistics, behavioral science, and optimization—not black-box ML; Analyze longitudinal behavior patterns to identify drivers, frictions, and causal levers; Apply time-series and probabilistic forecasting with uncertainty and scenario analysis; Design, evaluate, and implement LLM-based solutions — including RAG pipelines, classification, and extraction tasks — with rigorous benchmarking, calibration analysis, hallucination measurement, and bias auditing to ensure outputs are explainable; Act as a senior reviewer and methodological authority across data science initiatives; Set informal standards for rigor, documentation, and reproducibility; Mentor senior and mid-level data scientists through technical guidance and peer review; Translate complex quantitative results into clear, decision-oriented narratives for senior stakeholders; Challenge assumptions and narratives not supported by evidence; Influence strategy by grounding discussions in data, causality, and expected impact