New opportunity

Data Scientist

BNP Paribas Wealth Management · Singapore

About this role

The Data Scientist joins the AI for IT squad in BNP Paribas Wealth Management IT in Singapore. The role designs, evaluates, prototypes, implements and industrializes secure, governed AI capabilities that IT teams can use in their daily work. Its three focus areas are software delivery (Build), production operations (Run) and cybersecurity (Secure).

Responsibilities

Evaluate AI coding tools and prototype assistants for code generation and review, testing, documentation, refactoring, developer onboarding and retrieval of engineering knowledge.
Develop AIOps use cases for monitoring, alert triage, incident diagnosis, root cause analysis and runbook recommendations or execution support. Work with production and application teams to improve incident response.
Develop AI-assisted vulnerability analysis, prioritization and remediation while keeping security decisions traceable and subject to appropriate human validation.
Enhance specialized chatbots and agents and contribute to the WM GENIX platform. Build RAG capabilities, prompt patterns, evaluation datasets, knowledge bases and workflows; assess LLMs, agentic architectures, MCP integrations and tool calling. Current knowledge sources include SharePoint and Confluence, with further integrations expected.
Turn IT needs into specifications, user stories and measurable outcomes. Monitor adoption, productivity, quality, incident-resolution and remediation KPIs; support roadmap prioritization, technical and risk reviews, stakeholder collaboration, knowledge sharing and mentoring. Apply security, privacy, auditability, cost control and human oversight principles.

Requirements

Master’s degree in Computer Science, Data Science, Artificial Intelligence, Engineering, Mathematics, Statistics or a related field, and a minimum of 5 years’ experience in data science, AI engineering, software engineering or related technology roles.
Hands-on delivery of AI, GenAI, data science or automation solutions and experience in enterprise IT. Strong Python and AI engineering skills; understanding of LLMs, prompt engineering, RAG, embeddings, vector databases, AI assistants and agents.
Knowledge of software engineering and CI/CD workflows, AIOps, cybersecurity and vulnerability management, APIs, data pipelines, enterprise knowledge sources, and cloud and containerization concepts. The role calls for analytical problem-solving, practical delivery, communication, autonomy and accountability.

Preferred

Experience in banking or another regulated industry; AI for developer productivity, operations, cybersecurity or knowledge management; AI coding tools; integrations with enterprise collaboration or ITSM platforms; and Azure AI, cloud, cybersecurity or data science certifications. The position is permanent and full-time in Singapore.

For full details about this role, click “Apply on company site” to view the employer’s original posting.

Skills for this role

PythonData ScienceAI EngineeringGenerative AILarge Language ModelsPrompt EngineeringRetrieval-Augmented GenerationEmbeddingsVector DatabasesAI AgentsTool CallingModel Context ProtocolAIOpsCybersecurityVulnerability ManagementSoftware Development Life CycleCI/CDAPIsData PipelinesDockerKubernetesSharePointConfluenceJiraServiceNowMicrosoft 365CursorGitHub CopilotIncident TriageRoot Cause Analysis

Checking your profile…

Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.

Understand the role

What matters most

Improve software delivery with AI
The role evaluates AI coding tools and prototypes support for code generation, review, testing, documentation, refactoring and engineering knowledge retrieval. Prepare to explain how you would assess workflow value and output quality, not just generate a demo.
Support production operations
AIOps work covers monitoring, alert triage, incident diagnosis, root cause analysis and runbook support, in collaboration with production and application teams. Prepare an approach to identifying operational pain points and measuring incident-response improvement.
Advance governed vulnerability management
The role applies AI to vulnerability analysis, prioritization and remediation recommendations while keeping security decisions traceable and validated by appropriate owners. Prepare to distinguish AI recommendations from accountable security decisions.
Industrialize assistants and WM GENIX
The posting calls for improving existing chatbots and agents, RAG, evaluation datasets, knowledge bases and workflows, including assistants connected to SharePoint and Confluence. Prepare to discuss reliability evaluation and the path from prototype to usable enterprise capability.
Prioritize and prove value across Build, Run and Secure
The role turns IT needs into specifications and measurable outcomes, helps prioritize by value, feasibility, risk and readiness, and monitors adoption, quality, productivity and operational KPIs. Prepare to show how you would select use cases and define evidence of impact.

Likely challenges to think through

Developer speed versus code quality
Likely challenge, not a reported problem: a coding assistant may save time while introducing defects or weak tests. Prepare a comparison method that checks review and test quality alongside productivity before recommending a tool.
Less alert noise versus missed incidents
Likely challenge, not a reported problem: aggressive AI-assisted triage could suppress signals production teams need. Prepare to discuss evaluation against incident outcomes, escalation boundaries and human review of consequential recommendations.
Faster remediation versus accountable decisions
Likely challenge, not a reported problem: AI can accelerate vulnerability prioritization, but recommendations still need traceability and owner validation. Prepare to explain what evidence a recommendation should carry and where approval remains with security owners.
Broader knowledge access versus governed retrieval
Likely challenge, not a reported problem: connecting assistants to more enterprise sources may improve answers while complicating privacy, access control and answer reliability. Prepare an evaluation approach for source permissions, retrieval quality and auditable responses.

Practice your answers

Technical practice

How would you evaluate AI coding tools for an IT development team?

What to coverThe posting calls for assessing coding environments and their effects on code generation, review, testing and documentation. Suggested preparation: compare output quality, developer adoption, security controls and cost using representative tasks rather than relying on a polished demo.

How would you prototype an AI-assisted SDLC workflow without weakening release controls?

What to coverThe role includes test generation, documentation assistance and development workflow automation. Suggested preparation: explain where human review, automated tests and CI/CD checks would sit, and how you would detect low-quality generated changes.

How would you design and assess AI-assisted incident triage?

What to coverThe posting names monitoring, alert correlation, diagnosis, root cause analysis and runbook support. Suggested preparation: describe how to reduce alert noise while preserving evidence, distinguishing recommendations from actions, and measuring incident-resolution improvement.

How would you prioritize vulnerabilities with AI while keeping security owners accountable?

What to coverThe role covers vulnerability analysis, remediation recommendations and cyber backlog optimization, with traceable decisions validated by appropriate owners. Suggested preparation: discuss risk-based prioritization, the evidence behind recommendations and the point at which a human must approve a decision.

How would you build RAG over SharePoint and Confluence content?

What to coverThe posting identifies these as current knowledge sources for specialized assistants and calls for embeddings, vector databases and knowledge-base structures. Suggested preparation: explain ingestion, content freshness, access permissions, retrieval quality and how an assistant should respond when evidence is insufficient.

How would you evaluate and improve a specialized GenAI assistant?

What to coverThe role calls for prompt patterns, evaluation datasets and improvements to assistant quality, reliability and adoption. Suggested preparation: propose representative test cases and measures for grounded answers, error rates and user usefulness, then explain how findings would guide prompt or retrieval changes.

When would you use an agent with tool calling rather than a simpler assistant?

What to coverThe posting calls for prototyping agentic architectures, MCP-based integrations and tool-calling patterns. Suggested preparation: compare flexibility against complexity, cost and operational risk, including limits on tool permissions and human oversight for consequential actions.

How would you integrate an assistant with enterprise knowledge and workflow systems?

What to coverThe role involves APIs, data pipelines and enterprise sources; Jira, ServiceNow, Clarity and Microsoft 365 are described as expected future integrations, not established current sources. Suggested preparation: discuss authentication, permission-aware access, data synchronization, failure handling and auditability.

What would it take to move a Python AI prototype into governed production use?

What to coverThe posting emphasizes hands-on Python delivery, industrialization, CI/CD, deployment concepts, privacy and cost control. Suggested preparation: outline testing, monitoring, deployment and rollback, alongside technical and risk-review evidence needed before wider use.

How would you turn an IT pain point into a measurable AI use case?

What to coverThe role includes specifications, user stories, evaluation criteria and KPIs across Build, Run and Secure. Suggested preparation: define a baseline and success measure, then weigh value, feasibility, risk and readiness when deciding whether to advance a prototype.

Behavioural practice

Tell me about a time you aligned teams with different priorities on a technical solution.

What to coverThe role calls for collaboration across IT, production, architecture and cybersecurity stakeholders. Use a real situation, your actions to clarify concerns and reach a decision, and the resulting outcome.

Describe a time you moved an exploratory idea toward dependable day-to-day use.

What to coverThe posting emphasizes ownership from analysis through prototyping and industrialization. Use a real situation, the actions you took to validate and improve the solution, and the result, including any limits you identified.

Tell me about a time you had to prioritize competing technical initiatives.

What to coverThe role supports roadmap choices based on value, feasibility, risk and readiness. Describe the situation, how you gathered evidence and explained the trade-offs, and what resulted.

Describe a time you raised a security, privacy or governance concern about a proposed solution.

What to coverThe posting stresses responsible AI, traceability and human accountability. Give a real situation, the actions you took with the relevant owners, and the outcome without implying that risk could be eliminated entirely.

Tell me about a time you helped a junior colleague structure technical work.

What to coverThe role includes mentoring junior profiles or interns and contributing to knowledge sharing. Describe the situation, how you set direction and provided feedback, and the result for the work and the colleague.

Sharpen your knowledge

Topics to prepare

RAG and enterprise knowledge
Posting: The role builds RAG capabilities for assistants using knowledge sources including SharePoint and Confluence. Suggested revision: sketch how documents move through ingestion, embeddings, retrieval and answer generation, then explain how you would test answer quality.
Agents and controlled tool use
Posting: The role evaluates agentic architectures, MCP-based integrations and tool calling. Suggested revision: diagram an assistant that retrieves knowledge and proposes an action; mark where permissions, logging and human approval belong.
AI in software delivery
Posting: Build work covers AI coding tools, code review, unit tests, documentation and CI/CD workflows. Suggested revision: compare two coding-assistant use cases on correctness, review effort and risks before recommending one.
AIOps and incident response
Posting: Run work includes alert correlation, triage, diagnosis, root cause analysis and runbook support. Suggested revision: walk through a hypothetical alert cluster and distinguish evidence for a likely cause from a suggested response.
Risk-based vulnerability management
Posting: Secure work covers vulnerability prioritisation and remediation, with traceable decisions validated by appropriate owners. Suggested revision: explain how you would rank three hypothetical findings and record the rationale and approval point.
Governance and value measurement
Posting: Solutions must address privacy, auditability, cost and human oversight; the role defines adoption, quality, incident-resolution and remediation KPIs. Suggested revision: draft a prototype-to-production review checklist and choose one baseline and outcome measure for a proposed use case.

Skills to revise

Prototype a Python RAG workflow
Posting: Strong Python skills and hands-on RAG delivery are called for. Suggested exercise: use sample engineering documents to implement ingestion, retrieval and cited answers, then test failures with a small evaluation set. This mirrors the assistant and knowledge-base work.
Evaluate an AI coding assistant
Posting: The role assesses AI development tools, including Cursor and Kline/Cline-style tools. Suggested exercise: apply an available assistant to a small code-change and unit-test task; document correctness, review effort and security concerns. Specific tool experience is described as good to have.
Triage a simulated incident
Posting: The role prototypes alert triage, diagnosis and runbook recommendations with production teams. Suggested exercise: group sample alerts, write an evidence-backed incident summary and propose a runbook step for human review.
Document a vulnerability decision
Posting: AI-assisted remediation must remain governed and traceable. Suggested exercise: turn sample vulnerability records into a prioritised backlog with evidence, proposed remediation and an explicit owner-validation step.
Specify and measure an IT use case
Posting: The role turns IT pain points into specifications, user stories, evaluation criteria and measurable outcomes. Suggested exercise: write one user story for a support assistant, define an acceptance test and select adoption and resolution-quality measures.

Your preparation checklist

Map the stated requirements to evidence
Prepare a one-page evidence map for the stated master’s degree, five years of relevant experience, enterprise IT delivery, Python, AI engineering and integrations. Distinguish required experience from the listed advantages and certifications.
Rehearse a delivered-project walkthrough
Choose an actual AI or automation project, if available. Prepare a concise account of the IT problem, your contribution, architecture, evaluation, deployment decisions and measured results; identify any parts that remained prototypes.
Compare AI coding approaches
Draft a small evaluation matrix for a coding assistant in the Build domain. Use code review or test generation as an example, with criteria for output quality, developer effort, repository access, security and fit with CI/CD.
Sketch an incident-support workflow
Prepare a hypothetical Run-domain design showing how alerts and operational context could support triage, diagnosis and runbook recommendations. Specify evaluation against past incidents and where production staff would approve actions.
Work through a vulnerability decision
Prepare a hypothetical Secure-domain example that moves from vulnerability findings to prioritization and remediation recommendations. Show the evidence retained for traceability and the point at which an appropriate security owner validates the decision.
Design a governed knowledge assistant
Sketch a RAG design relevant to the role’s WM GENIX work, using SharePoint or Confluence as an example source. Cover ingestion, permissions, retrieval, prompt patterns, evaluation data and safeguards for tool calling or agents; label the design as proposed, not an existing implementation.
Plan the path from prototype to adoption
Draft a sample user story and success measures for one of your designs. Include quality, adoption, cost and an appropriate productivity, incident-resolution or remediation KPI, plus the security, privacy and risk questions to resolve before production.
Prepare questions for the AI for IT team
Ask how Build, Run and Secure initiatives are prioritized; what is already deployed versus exploratory on WM GENIX; which knowledge-source integrations are approved; and how teams assess value and human oversight.

Availability last checked: 4 Nov 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy

KEEP EXPLORING

Similar AI jobs

Related skills and specializations in Singapore. Matched to this role, not your profile.

Explore more jobs

Finding similar opportunities…