New opportunity

Forward Deployed (AI) Engineering Lead – Global Broking, APAC

TP ICAP · Singapore

About this role

The Forward Deployed (AI) Engineering Lead will identify, design and deliver AI-enabled productivity improvements for TP ICAP’s Global Broking desks across APAC. Based in Singapore in a hybrid role, the lead works directly with brokers, desk heads and business managers to reduce manual work, improve information discovery and support faster decisions. The position is accountable to APAC Global Broking leadership for business outcomes and also reports to Global AI Lab / Technology leadership for technical and governance alignment. Band and level: Manager, 7.

Responsibilities

Understand front-office workflows and desk pain points; translate business needs into a prioritised AI opportunity pipeline with expected benefits, dependencies, delivery status and business owners.
Design, prototype and deliver AI tools, agents, applications and workflow automations. Potential use cases include broker preparation, market colour aggregation, document automation, research support, data extraction, workflow orchestration and cross-system information retrieval. Use approved platforms, including AWS Quick, AWS Kiro, Microsoft 365 / Copilot and internal AI platforms.
Lead use cases from discovery through production deployment and desk adoption. Define measurable outcomes and adoption plans; support pilots, workshops, hackathons and AI enablement. Turn successful pilots into scalable opportunities and create reusable delivery patterns and documentation.
Apply architecture, code management, testing, production-readiness and operational-support practices. Coordinate with Technology, Architecture, Security, Risk and Compliance on AI risk assessments, reviews, approvals and controls; escalate blockers through business and technology reporting lines.

Required background:

Proven experience in software engineering, AI engineering, automation, data engineering or technology delivery in financial services or another regulated environment. Practical experience delivering solutions with large language models, AI agents, workflow automation, RAG or knowledge retrieval, and API integrations.
Ability to work with front-office or business users; knowledge of technology governance, security-by-design, deployment and support; experience with Python, APIs, cloud platforms, source control, CI/CD and structured documentation. Strong communication and stakeholder-management skills are required. No minimum years or education requirement is stated.

Advantageous experience includes AWS Bedrock, AWS Quick, Microsoft Copilot / M365 integration, Power Platform, Teams, SharePoint, data extraction or workflow automation platforms, and knowledge of interdealer broking, capital markets, wholesale financial markets or front-office operating models.

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

Skills for this role

AI engineeringSoftware engineeringData engineeringLarge language modelsAI agentsRetrieval-augmented generationKnowledge retrievalWorkflow automationAPI integrationPythonCloud platformsSource controlCI/CDAWS QuickAWS KiroAWS BedrockMicrosoft 365Microsoft CopilotPower PlatformMicrosoft TeamsSharePointData extractionArchitecture reviewSecurity by designTestingProduction deploymentTechnology governanceRisk managementStakeholder managementFront-office workflows

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

Build a business-owned AI opportunity pipeline
Work with APAC brokers, desk heads and managers to understand front-office workflows and rank opportunities by expected benefits, dependencies and business ownership.
Deliver practical AI solutions
Design and build tools, agents and automations on approved platforms. The role names broker preparation, market colour aggregation, document automation and cross-system retrieval as possible use cases.
Turn deployments into measurable desk adoption
Lead use cases from discovery through production and day-to-day use, with defined benefit measures, desk sponsors and adoption plans.
Engineer for governed production use
Apply architecture, code management, testing, documentation and operational-support practices; coordinate risk assessments, reviews and approvals with Technology and control functions.
Coordinate and reuse delivery across teams
Serve as the APAC Global Broking AI engineering lead between desks, the AI Lab and support functions. The stated success measures include converting pilots into scalable opportunities and sharing reusable patterns, documentation and controls.

Likely challenges to think through

Desk fit versus broader reuse
A likely trade-off is meeting desk-specific workflow needs without creating solutions that cannot scale across APAC. Prepare a way to distinguish local requirements from reusable components, and to record benefits and ownership before prioritising.
Prototype speed versus production controls
Rapid AI experimentation may be difficult to reconcile with security, data, risk and production requirements. Prepare an example approach that introduces testing, documentation, review and support planning early rather than after a pilot.
Business outcomes versus technical alignment
The dual reporting model could make priority and dependency decisions complex; this is a preparation hypothesis, not a stated problem. Prepare to explain how you would make trade-offs visible to business and Technology leaders and escalate blockers through both lines.

Practice your answers

Technical practice

How would you identify and prioritise AI opportunities across APAC broking desks?

What to coverSuggested answer focus: map repetitive tasks, information flows and desk-specific pain points with brokers and desk heads. Explain how you would rank opportunities by expected benefit, feasibility, dependencies and risk, while recording business owners and delivery status in the pipeline described in the posting.

How would you turn a broker-preparation need into a production-ready AI solution?

What to coverSuggested answer focus: define the user workflow and success measure before choosing an AI tool, application or automation. Walk through prototyping, architecture and security review, testing, production support and desk adoption; these stages reflect the posted responsibilities.

How would you design and evaluate cross-system knowledge retrieval for a broking desk?

What to coverSuggested answer focus: discuss source selection, access permissions, retrieval quality and how an LLM would ground its responses in retrieved material. As a preparation suggestion, describe tests for missing or conflicting information and when the user should check the underlying source.

When would you use an AI agent rather than a simpler workflow automation?

What to coverSuggested answer focus: compare flexible, multi-step reasoning with a more predictable automation for a stated task such as research support or workflow orchestration. Address testing, oversight and operational complexity; the posting calls for both agents and workflow automations, not an agent for every use case.

How would you integrate a Python-based AI application with existing systems through APIs?

What to coverSuggested answer focus: explain data contracts, authentication, error handling and how the integration would respect access controls. As a preparation suggestion, describe how you would test failures and document dependencies so the application can be supported after deployment.

How would you select an approved platform for an AI use case?

What to coverSuggested answer focus: compare fit with the workflow, integration needs, security requirements and support model before choosing among approved environments such as AWS Quick, AWS Kiro, Microsoft 365 / Copilot or internal AI platforms. Distinguish these named platforms from AWS Bedrock and Power Platform, which the posting lists as advantageous experience.

How would you assess a document-automation or data-extraction prototype?

What to coverSuggested answer focus: define the fields or output the desk needs, then test accuracy, completeness and handling of ambiguous inputs. As a preparation suggestion, discuss when human review would be appropriate and whether the time saved outweighs correction effort.

What engineering practices would you require before deploying a desk-facing AI tool?

What to coverSuggested answer focus: cover architecture, source control, CI/CD, structured documentation, testing, production-readiness checks and an operational-support arrangement. Explain how you would make failures visible and resolve them without disrupting the desk workflow.

How would you handle AI risk and architecture approvals without losing delivery momentum?

What to coverSuggested answer focus: identify security, data, risk and compliance considerations early and keep Technology and control stakeholders informed throughout delivery. Describe what you would document for assessments and reviews, and when you would escalate a blocker through the business and technology reporting lines.

How would you prove that a pilot delivered value and decide whether to scale it?

What to coverSuggested answer focus: establish a baseline, define a productivity measure and gather evidence of desk adoption with the business sponsor. Explain how pilot feedback, support needs and reusable documentation would inform broader APAC or global rollout, as described in the success measures.

Behavioural practice

Tell me about a time business and technology stakeholders wanted different priorities.

What to coverSuggested answer focus: use a real situation, the actions you took to clarify trade-offs and ownership, and the result. Relate your example to the role’s dual accountability for broking outcomes and technology alignment, without claiming experience you do not have.

Describe a time you earned the trust of business users while changing their workflow.

What to coverSuggested answer focus: set out the users’ initial concern, how you learned their day-to-day work and involved them in testing, and what changed as a result. The posting emphasises direct work with brokers and desk heads and adoption of usable solutions.

Tell me about a time you slowed or changed a delivery to address a risk.

What to coverSuggested answer focus: describe the situation, your judgement and communication with relevant stakeholders, and the outcome. Choose an example that shows how you balanced delivery speed with security, governance or production discipline, rather than treating controls as an afterthought.

Describe a pilot that needed substantial changes before users adopted it.

What to coverSuggested answer focus: explain the original objective, what feedback or usage evidence revealed, the changes you made and the result. The role calls for measurable benefits and for turning useful pilots into scalable opportunities.

Tell me about a blocker you resolved across teams or reporting lines.

What to coverSuggested answer focus: give the situation, your actions to make the dependency visible and coordinate the right people, and the result. The posting calls for prompt escalation and collaboration across desks, the AI Lab, Technology and control functions.

Sharpen your knowledge

Topics to prepare

Front-office workflows and opportunity prioritisation
The role requires understanding broker workflows and maintaining a prioritised APAC AI opportunity pipeline. Revise by mapping a hypothetical desk’s information flow and manual steps, then ranking three opportunities by expected benefit, dependencies and business ownership.
LLMs, agents and retrieval design
The posting calls for practical delivery using LLMs, AI agents, RAG or knowledge retrieval, workflow automation and API integrations. Revise by sketching when a broker-preparation task needs retrieval, an agent or simpler automation, and how you would check the output against its sources.
Governance and security-by-design
The lead must align solutions with architecture, security, data, risk and compliance requirements. Revise by walking through a hypothetical cross-system retrieval design: identify data-access risks, proposed controls, review stakeholders and points requiring approval.
Production delivery, adoption and benefits
The role spans discovery through production deployment and desk adoption, with measurable productivity outcomes. Revise by defining a pilot’s baseline measure, adoption measure, production-readiness criteria and support arrangement, then explaining how evidence would justify scaling it.
Broking and wholesale-market context
Knowledge of interdealer broking, capital markets, wholesale financial markets or front-office operating models is advantageous, not required. As optional preparation, trace a hypothetical broker’s preparation and market-colour workflow and identify where faster information discovery might help.

Skills to revise

Run desk discovery and build a use-case pipeline
Practise a short discovery interview with a hypothetical desk head. Turn the answers into a one-page pipeline showing the pain point, sponsor, expected benefit, dependencies, delivery status and priority; these are explicit responsibilities in the posting.
Prototype a Python retrieval and API workflow
Using synthetic documents, build a small Python prototype that retrieves relevant passages and passes them to an LLM through an API; test unsupported and missing-source answers. This exercises the stated Python, API, LLM and knowledge-retrieval experience against the advertised research-support and information-discovery use cases.
Prepare a production-ready delivery package
For that prototype, practise source control, a CI/CD check, tests, architecture notes and an operational-support handover. The posting explicitly expects code management, testing, documentation, deployment and production-readiness discipline.
Document an AI review and escalation
Draft a concise risk assessment and solution summary for the prototype, listing data access, security controls, architectural questions, approval needs and unresolved blockers. Practise explaining what you would escalate to business leadership versus AI Lab / Technology; both reporting lines and control-function reviews are stated.
Present a pilot and adoption plan
Rehearse a five-minute briefing for a desk sponsor and technical stakeholders: demo the workflow, state a measurable productivity outcome, propose a pilot and adoption plan, and explain how feedback would inform rollout. The role explicitly involves stakeholder engagement, enablement and benefits realisation.
Explain approved-platform choices
Practise describing where AWS Quick, AWS Kiro, Microsoft 365 / Copilot or internal AI platforms might fit a proposed use case, while flagging any capability or access you would need to verify. The posting names these as approved platforms; hands-on familiarity with AWS Bedrock and specific M365 integrations is described as advantageous.

Your preparation checklist

Map a desk workflow and rank opportunities
The role calls for a prioritised APAC AI pipeline. Prepare a hypothetical broker-workflow map and a short opportunity table showing manual effort, expected benefit, dependencies, business owner and delivery status; label assumptions for employer validation.
Build a relevant delivery walkthrough
Select an actual project you can discuss and prepare a one-page account of the user problem, your contribution, AI or automation approach, deployment, measured outcome and lessons learned. The posting seeks delivery experience in regulated environments and work with business users; do not imply your project was at TP ICAP.
Practise an AI solution design exercise
Sketch a proposed broker-preparation or cross-system retrieval tool, both use cases named in the posting. Show data and API flows, a possible LLM or RAG component, testing, failure handling and a path from prototype to supported production; treat platform choices as conditional on approval.
Assemble engineering and governance evidence
Prepare concrete examples of Python and API work, cloud deployment, source control, CI/CD and structured documentation. Pair them with an example of security-by-design, architecture review, AI risk assessment or production-readiness work, distinguishing what you did from what others approved.
Plan how to prove adoption and benefits
Draft a sample pilot scorecard with a baseline, productivity measure, usage or adoption signal, desk sponsor and decision point for scaling. The posting explicitly calls for measurable benefits, desk adoption and reusable delivery patterns; the scorecard is a preparation exercise, not a stated TP ICAP template.
Rehearse cross-functional leadership decisions
Prepare a concise example of balancing a business delivery priority against a technology or control concern, including how you communicated and escalated a blocker. The role has accountability to APAC Global Broking leadership and a reporting line to Global AI Lab / Technology leadership.
Bring questions about the operating model
Write questions for the employer about initial desk priorities, how opportunities are sponsored, which platforms are approved for a given use case, how risk and production approvals work, and what evidence turns an AWS Quick or other AI pilot into a scalable rollout.

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…