Director Data Science

JOB SUMMARY

The Director of Data Science leads BSWH's enterprise data science and advanced analytics portfolio. The role converts business, clinical, customer, and operational priorities into measurable analytical products, including predictive, prescriptive, optimization, experimentation, machine learning, and AI capabilities. The Director is accountable for analytical approach selection, model and experiment evaluation, transition to production with AI Engineering, post-launch monitoring, adoption, and evidence of value.

Within the Data & Knowledge function, this leader establishes the data science operating model, prioritizes investment and capacity, and builds a high-performing team. The Director maintains a portfolio of opportunities with clear value hypotheses, success measures, dependencies, risks, and decision points, and recommends when to build, buy, partner, defer, or retire a capability.

This role works closely with Enterprise Data, Ontology & Knowledge Products, AI Engineering, Product, Digital, Clinical, Operations, Governance, Privacy, Security, and other partners. These partners retain accountability for their respective product, platform, source-data, policy, and clinical approval decisions. Where data science supports clinical or patient-facing decisions, the Director ensures that the intended use is appropriately validated, governed, and implemented with clinical and operational leaders.

This position can be based in our administrative building in Dallas, Texas or mostly remote with some travel required.

ROLE ACCOUNTABILITIES

Portfolio Strategy and Prioritization
- Define and execute the enterprise data science roadmap in alignment with BSWH strategic priorities, the AI use-case portfolio, and measurable value opportunities.
- Establish a consistent intake and prioritization process that assesses expected value, feasibility, data readiness, risk, dependencies, capacity, and time to impact.
- Maintain a portfolio view for major initiatives, including the problem owner, value hypothesis, success measures, stage, decision points, dependencies, and retirement criteria.
- Recommend staffing, funding, vendor, and build-or-buy decisions for data science capabilities and analytical products.
- Set standards for model development, experimentation, documentation, validation, monitoring, and lifecycle management.

Analytical Methods, Machine Learning and AI Enablement
- Select statistical, machine learning, optimization, simulation, GenAI, or agentic AI methods that fit the problem, intended use, data quality, risk, and healthcare context.
- Lead the development of predictive, prescriptive, forecasting, risk, recommendation, personalization, propensity, optimization, and decision-support capabilities across priority domains.
- Guide analytical design and evaluation for generative and agentic AI use cases, including task success, groundedness, safety, human escalation, and operational fit where applicable.
- Partner with AI Engineering and Product teams to define interfaces, deployment requirements, monitoring expectations, service levels, and handoffs for production use.
- Ensure analytical capabilities are reusable, interpretable where appropriate, maintainable, and designed with clear human oversight and intended-use limitations.

Model Evaluation, Experimentation and Value Measurement
- Define pre-deployment and post-deployment evaluation frameworks that cover technical quality, business value, statistical validity, operational impact, fairness and equity, reliability, safety, and customer or clinical outcomes.
- Lead test design, measurement strategy, causal inference, A/B testing, experimentation, and impact evaluation where appropriate.
- Establish success criteria and monitoring measures for model performance, subgroup performance, calibration, bias, drift, adoption, cost, latency, and workflow fit.
- Define requirements for model documentation, model inventory, validation evidence, monitoring thresholds, incident escalation, change control, and retirement.
- Require each major initiative to have a value hypothesis, adoption plan, outcome measures, and a post-launch review or value achievement plan.
- Partner with clinical leaders and appropriate review bodies when models affect care delivery, clinical decisions, patient safety, or…