Senior Vice President, Data Science Manager

What matters most

• Governed proposal knowledge base

  The role leads a scalable content ecosystem for proposals, RFPs and due diligence. It includes integrating structured and unstructured sources and setting taxonomies, metadata, editorial standards, approvals and version control.

• Retrieval and answer recommendation

  The role applies NLP and language-focused machine learning to help teams find and reuse approved content. Ranking must account for relevance, freshness, quality and reusability, not just textual similarity.

• Controlled AI-assisted drafting

  The posting calls for retrieval-augmented generation concepts, prompt design, response evaluation and human review. Drafts must be assessed for factual consistency, technical precision, tone and alignment with approved messaging.

• Cross-functional delivery

  This SVP role combines hands-on execution with strategic leadership. It partners with sales, product, solutions, deal management, controls and subject matter experts, and leads work from problem definition through delivery.

• Measurable proposal improvement

  The stated aim is better, faster and more complete first drafts with less manual work and unnecessary expert outreach. The role defines measures including reuse, answer acceptance, manual edits, SME touch frequency and cycle time.

Likely challenges to think through

• Finding content without resurfacing stale language

  Preparation hypothesis: a highly relevant passage may still be outdated or lack the right approval status. Prepare to explain how lifecycle metadata, version control and ranking criteria could keep retrieval useful while respecting governance.

• Making drafting faster without weakening controls

  Preparation hypothesis: AI-assisted first drafts could reduce effort but introduce imprecise control, risk or service claims. Prepare an evaluation and human-review approach that distinguishes acceptable reuse from content requiring expert validation.

• Reducing SME outreach while preserving expertise

  Preparation hypothesis: capturing expert contributions as reusable assets takes coordination, even if it lowers repeat requests later. Prepare to discuss how content gaps could be prioritized with stakeholders and how SME touch frequency and answer acceptance would show whether the approach helps.

Topics to prepare

• Content governance and lifecycle

  The posting requires taxonomies, ontologies, metadata, approvals and version control for reusable proposal content. Revise by sketching how a hypothetical approved Q&A moves from expert input through review, publication, revision and retirement.

• Retrieval and answer-ranking principles

  The role calls for semantic search and recommendations based on relevance, freshness, quality, specificity and reusability. Revise by comparing how those factors would rank two plausible answers to the same due diligence question.

• Controlled AI-assisted drafting

  The posting names retrieval-augmented generation, prompt design, human review and responsible AI. Revise by outlining when a draft should cite approved content, request review or decline to make an unsupported claim.

• Language quality and control sensitivity

  The role must balance technical and service-related precision with clear client-facing language. Revise by editing a sample service description for factual consistency, clarity, tone and alignment with approved wording.

• Proposal workflow and impact measures

  The posting specifies reuse, answer acceptance, first-draft quality, manual edits, SME touch frequency and cycle time. Revise by defining each measure and explaining how to distinguish a faster draft from a better one.

Skills to revise

• Python text analysis

  Python and NLP libraries are required. Practice cleaning a small mock Q&A corpus, generating sentence embeddings, finding near-duplicates and inspecting classification errors; relate the results to content reuse.

• SQL for content and usage analysis

  Strong SQL is required for text-centric workflows. Practice joining mock content versions, approval status and usage records to find stale, frequently reused answers and unanswered topics.

• Metadata and schema design

  Enterprise content organization is central to the role. Draft a sample schema for Q&A pairs and response modules, including topic, product or service, approval status, owner and version; test whether it supports retrieval and governance.

• Retrieval and draft evaluation

  The posting requires ranking frameworks and response evaluation. Using hypothetical approved passages, score candidate answers for relevance and freshness, then review a draft for unsupported statements and needed human approval.

• Stakeholder delivery

  Hands-on leadership across sales, product, controls and SMEs is required. Rehearse a brief problem-definition discussion that turns a content gap into scope, milestones, owners and a measurable outcome.

• Office-based communication

  High proficiency in Excel, PowerPoint and Word is required. Practice summarizing mock reuse and edit-rate data in Excel, presenting a concise recommendation in PowerPoint, and editing a controlled response in Word.

• Build a requirements evidence sheet

  Map the posting’s required experience to specific examples you can discuss: large unstructured-text repositories, Python and SQL, NLP, content governance, responsible AI, and hands-on leadership. Note any gaps rather than inventing evidence.

• Prepare a project walkthrough

  Choose a relevant project, if you have one, and outline the problem, corpus, methods, your decisions, stakeholder roles, results, and limitations. Emphasize how accuracy and governance affected delivery.

• Sketch a governed content model

  As an interview exercise, diagram how proposal Q&A, reusable modules, product descriptions, and approved artifacts could share taxonomy, metadata, approval status, ownership, and version history. Distinguish this proposed design from BNY’s actual systems.

• Practice a retrieval and drafting design exercise

  Outline how you would normalize sources, detect duplicates, retrieve and rank content for relevance and freshness, and support drafting with human review. Explain where controls would prevent outdated or unapproved language from being reused.

• Draft an evaluation scorecard

  Define how you would assess factual consistency, technical precision, clarity, tone, and alignment with approved messaging. Include proposed measures from the posting—such as answer acceptance, manual edits, SME contact frequency, and cycle time—and how you would establish a baseline.

• Prepare a delivery and stakeholder plan

  Describe a possible sequence from problem definition and content-gap analysis to a limited pilot and measured rollout. Be ready to explain how sales, product, deal management, controls, and SMEs would help validate content and resolve trade-offs.

• Bring focused questions for BNY

  Ask which content sources and approval boundaries are in scope, where first drafts most often need correction, who owns content freshness, and how the team currently measures reuse and SME outreach.