Senior Associate/Assistant Vice President, AI Product Manager

Introduction

AI product management is a distinct discipline from conventional software product management. The AI PM must navigate probabilistic, non-deterministic systems; must be fluent enough technically to shape model selection, evaluation strategy, and prompt architecture; and must be skilled enough in business analysis to translate ambiguous enterprise problems into AI product specifications that engineering teams can actually execute.

The AI Product Manager at Temasek owns the end-to-end product lifecycle for one or more internal AI products — from problem framing and business case through to production deployment, adoption, and continuous improvement. You will work at the intersection of investment workflows, agentic AI platform capabilities, and user experience — defining what to build, why it matters, and how success is measured. You will not throw requirements over a wall to engineering; you will be an active, technically-engaged product leader embedded in cross-functional delivery teams.

Responsibilities

AI-native product ownership

• Own the full lifecycle of assigned AI products — from problem discovery and opportunity sizing through solution design, prioritisation, launch, adoption tracking, and continuous improvement — with clear accountability for measurable business outcomes.
• Define product vision and strategy for AI use cases across investment, portfolio monitoring, research, and operations workflows, translating senior stakeholder intent into clear roadmaps with sequenced milestones.
• Make informed decisions on AI system design, including selection of approaches (e.g., RAG, fine-tuning, prompt engineering), structuring reliable agent workflows, incorporating human-in-the-loop checkpoints, and developing evaluation frameworks that reflect real user value.
• Drive rigorous product discovery by engaging investment professionals, shadowing workflows, analysing pain points, and synthesising insights into prioritised backlogs with clear rationale.
• Own the product evaluation strategy — defining success metrics, designing evaluation datasets, and maintaining benchmarks to track performance across model and prompt iterations.

Business-to-AI translation

• Translate complex and often implicit business needs into technically actionable AI product specifications, including data requirements, model capabilities, system integrations, and acceptable levels of uncertainty.
• Develop and maintain a structured product requirements framework tailored for AI systems, covering prompts and agent behaviour, tool and API definitions, evaluation rubrics, guardrails, and feedback mechanisms.
• Act as the primary interface between business stakeholders and engineering teams — managing expectations of AI capabilities, communicating probabilistic outputs clearly, and building trust through transparent performance reporting.
• Establish a clear business case framework linking product metrics (e.g., usage, accuracy, latency) to business outcomes (e.g., time saved, decision quality, risk reduction) to support ongoing investment decisions.

Cross-functional execution 

• Lead cross-functional delivery squads comprising AI engineers, data engineers, UX designers, and domain specialists — maintaining delivery momentum, managing inter-team dependencies, and escalating blockers with clear context and proposed solutions.
• Partner with the AI Security & Governance Lead to ensure products meet governance requirements at every stage — embedding compliance into the product development process rather than treating it as a launch gate. 
• Collaborate with the AI Fluency and Change Management team to design adoption programmes for launched AI products — defining the user education, rollout strategy, and feedback loops that convert deployment into sustained usage. 
• Engage regularly with the platform engineering team to feed product-level requirements into platform capability planning — ensuring that common patterns across products are generalised into reusable platform components rather than rebuilt product-by-product. 

Enterprise citizen developer enablement 

• Design and deliver structured enablement programmes that equip non-technical employees to build low-code, no-code, and AI-assisted solutions using Temasek's approved enterprise platforms (e.g. Microsoft Power Platform, Copilot Studio, or equivalent AI tooling). 
• Develop role-specific learning pathways for key employee segments — investment professionals, portfolio monitoring teams, operations, and support functions — that contextualize AI capability building within real workflow scenarios. 
• Build and maintain a library of pre-validated citizen developer starter templates, prompt patterns, and agent workflow blueprints that teams can adapt without deep technical expertise. 
• Run hands-on workshops, office hours, and embedded team sessions to accelerate capability from awareness to active production use. 

Requirements

Experience and background

• 4–8 years of product management experience, with at least 2 years owning AI or ML products from problem framing through production deployment — ideally in a fast-moving AI-native organisation (OpenAI, Anthropic, Google DeepMind, or equivalent) or a financial institution with a serious AI product function.
• Demonstrated ability to ship AI products that real users adopt and depend on — not just prototypes or proofs of concept, but production systems with measurable impact and ongoing improvement cycles.
• Sufficient technical depth to engage substantively with AI engineers: able to review and critique prompt architecture, evaluate RAG pipeline design choices, understand the implications of model selection, and participate in evaluation framework design.
• Domain experience in financial services — investment research, portfolio management, operations, or risk — is a strong advantage given the need to deeply understand the workflows this role will be building for.

Skills and capabilities

• Structured problem framer: able to take ambiguous business problems and decompose them into clearly scoped product opportunities with defined success criteria and realistic scope boundaries.
• Evaluation-minded: understands that AI product quality is measured differently from conventional software quality, and has experience designing and operating LLM evaluation frameworks (automated evals, human eval panels, red-teaming).
• Stakeholder translator: equally comfortable in a technical architecture discussion with engineers and a strategic discussion with investment professionals — adjusts vocabulary and framing without losing precision.
• Outcome-obsessed: tracks leading indicators (activation, usage depth, task completion) and lagging indicators (business impact, user confidence, error rate trends) as naturally as breathing — does not conflate shipping features with delivering value.