Senior AI Scientist - Smart Leads & Agentic Workflows (Business & Wealth)

• Sydney Based Location / Hybrid Working

The Opportunity

Westpac's Business & Wealth team is looking for a hands-on Senior AI Scientist to deliver smart leads models and agentic workflows that improve customer relevance, banker productivity and measurable value.

In this highly visible role, you will own complex use cases end to end, moving from a clearly framed business problem and robust data assessment through modelling, evaluation, implementation and adoption. Your role brings predictive customer insight together with AI-enabled workflows so bankers and relationship teams can prepare, prioritise and follow up opportunities more effectively, within well-defined action boundaries.

importantly, you thrive on staying close to the models, experiments and workflows; you can explain trade-offs without the jargon; and you care as much about fair customer outcomes, safe operation and adoption as you do about technical performance.

Responsibilities …

• Translate business priorities into smart leads and workflow use cases with clear users, outcomes and success measures.
• Develop customer propensity, opportunity and lead-prioritisation models using approved data and relevant business context.
• Build agentic workflows that help teams prepare, prioritise and follow up opportunities within approved action boundaries.
• Evaluate lead quality, model performance and incremental impact, testing whether interventions improve customer outcomes.
• Test agent accuracy, grounding and task completion, with human review, safe fallback and controlled tool use built in.
• Partner with engineering and business teams to integrate models and agents into practical, supportable workflows.
• Apply privacy, security, Responsible AI and model-risk controls, including consent, suitability and customer-vulnerability considerations.
• Monitor adoption, performance and customer impacts, then refine solutions using feedback, experiments and production evidence.
• Review technical work, mentor practitioners, share reusable components and escalate unresolved delivery or control issues.

Essential ingredients for success

• Strong hands-on Data Science experience in customer analytics, propensity modelling, lead prioritisation or decision science.
• Evidence of taking models or AI-enabled workflows from problem framing into production use, monitoring and improvement.
• Practical capability in experimentation and incremental-impact measurement, not only model accuracy.
• Applied experience with GenAI, retrieval or agentic workflows, including grounding, evaluation, human review and controlled actions.
• Strong Python and SQL skills, plus the ability to partner effectively with AI engineering and technology teams.
• Sound judgement across Responsible AI, privacy, security, model risk, fairness and customer safeguards.
• Clear stakeholder communication and credibility to review work, mentor practitioners and resolve delivery complexity.

How success will show up …

• More relevant engagement, fair treatment and improved relationship outcomes.
• Stronger lead quality, incremental conversion and workflow task success.
• Stable models, safe agent actions and effective customer safeguards.
• Banker adoption, time saved and evidenced business benefits.
• Higher review quality, practitioner growth and reuse across the squad.

Does this sound like you?

To get started, simply click on the APPLY or APPLY NOW button.

Please note that application closing dates are subject to change so don’t delay putting your best foot forward!

We’re all about creating a supportive and inclusive community. We welcome everyone – no matter your age, gender, background, or abilities. We also provide added support to welcome our veterans, Indigenous Australians, and neurodiverse community.

If you need any adjustments during the recruitment process, you can find out more information and additional contact details by visiting the "People with Disability and/or needing Accessibility Requirements" page on our website.