Data Scientist, Specialist

Vanguard•Malvern, PA

2d•Hybrid

About The Position

As a Data Scientist, Specialist, you will pair deep technical expertise with strong business partnership to turn data into decisions that drive measurable outcomes. You don’t just build models: you will be working closely with stakeholders across product, operations, and leadership to translate ambiguous business needs into structured analytic approaches. You will own meaningful pieces of the modeling stack end to end: from problem framing and data wrangling through model development, evaluation, deployment, and the design of how a human ultimately consumes and acts on the output. Success in this role requires not only technical excellence, but also the ability to influence decisions, align stakeholders, and communicate insights clearly enough to drive action. You will serve as an analytics expert on cross-functional teams supporting large strategic initiatives, bringing rigor, clarity, and perspective to both technical and business discussions. This is a forward-looking role: we are hiring people who can do excellent classical data science and operate as trusted partners in shaping how AI drives business value at Vanguard.

Requirements

5+ years of applied data science / ML experience, including work that reached production or directly drove business decisions.
Bachelor's degree in Statistics, Applied Mathematics, Computer Science, Economics, Analytics, or a related quantitative field; graduate degree preferred, or an equivalent combination of training and demonstrated experience.
Strong programming and data-wrangling skills in Python and SQL; comfort accessing, transforming, and preparing large-scale data for modeling.
Solid grounding in statistical and machine learning methods, including model validation, and the judgment to choose the right method for the problem.
Experience working in cloud environments (AWS, Azure, or GCP) and with modern collaboration/version-control tooling (e.g., Git, Jira, Confluence).
Ability to communicate technical findings to non-technical partners and to work cross-functionally across business, engineering, and leadership.
Building for the Age of AI Beyond classical data science, we're looking for people who are fluent, or eager to become fluent, in the tools and patterns reshaping the field.
Strength in several of the following matters more than checking every box: GenAI / LLM application, including retrieval-augmented generation (RAG), embeddings and semantic search, and prompt design.
Agentic systems: designing, orchestrating, and debugging multi-step LLM/agent workflows that use tools and take actions, using frameworks such as LangChain / LangGraph or equivalents.
LLM evaluation and reliability: building eval harnesses, defining quality and guardrail metrics, and knowing how to make non-deterministic systems trustworthy.
Causal inference and uplift modeling: treatment-effect estimation, experimentation, and designing for "what changes the outcome," not just "what predicts it."
MLOps mindset: model deployment, monitoring, drift detection, and the discipline of keeping a live model honest.
Responsible AI in a regulated context: explainability, fairness, and governance awareness appropriate to financial services and to models that drive real-world actions.
AI-augmented working style: using AI coding and analysis assistants to move faster, while critically evaluating their output rather than trusting it by default.

Nice To Haves

Experience with recommendation, ranking, next-best-action, or other decision-support systems.
Familiarity with feature stores, real-time or near-real-time inference, and vector databases.
Exposure to big-data frameworks (Spark, etc.).
Experience applying analytics across a range of business domains.

Responsibilities

Lead communication and influence decisions. Translate complex findings into clear, actionable narratives tailored to different audiences; align stakeholders on trade-offs, risks, and recommended actions, and ensure insights result in real business decisions.
Build and validate models end to end. Develop predictive and prescriptive models on large-scale data: from feature engineering and data foraging through model selection, calibration, and validation.
Make insights actionable, not just accurate. Design how model output is surfaced to the people who use it: the explanation, the context, and the recommended action. Optimize for a human making a good decision, not just for a leaderboard metric.
Ship to production and keep it healthy. Partner with MLE and engineering to deploy models, then own monitoring for drift, degradation, data quality, and real-world performance against business outcomes.
Probe the business, then structure the problem. Engage stakeholders to understand processes and drivers, bring structure to ambiguous requests, and translate them into a defensible analytic approach.
Design and run experiments. Apply sound experimental and causal reasoning to measure impact and to distinguish what predicts an outcome from what changes it.
Communicate with clarity. Prepare and deliver insight presentations and recommendations; translate complex findings and their implications for business partners and leadership.
Build with AI as a force multiplier. Use modern AI tooling (coding assistants, LLM-based workflows) to accelerate your own development, prototyping, and analysis — with sound judgment about where these tools help and where they don't.
Help grow the practice. Serve as an analytics expert on cross-functional strategic initiatives, contribute to research and reusable methods, and help raise the bar for the broader Vanguard analytics community.