Description

AI Product Manager

Why this role exists

The AI Center needs a single accountable leader for portfolio strategy and industrialisation-turning experiments into adopted products with controlled run-costs, clear governance, and decision-grade evidence. This role operationalises a "single-door" direction: close the door on ad-hoc tools and drive consistent, scalable AI product delivery through a governed factory model.

Portfolio scope

A) AI use-cases moving through Innovation Board gates

• Recipe.AI, Chef.AI, Portfolio.AI, Marketing Translation.

B) AI Center enablement services and assets

• Innovation Board operating rhythm and gate criteria
• AI Lab Principles/Playbook (PoC vs pilot vs MVP vs BAU)
• AI Toolkit + Sandbox adoption (standard experimentation route)
• Ethics enablement
• FinOps/showback, cost governance, run-cost strategy

Outcomes you will be measured on

A) Roadmap, investment logic, and value clarity

• 12-18 month portfolio roadmap with sequencing, dependencies, business cases (ROI + adoption assumptions), and operating model per product.
• Clear prioritisation logic based on value, feasibility, risk, and run-cost.

B) Innovation Board runs as a product factory

• Board has a clear charter, RACI, cadence, scorecard (funnel volume, pass/fail rates, cycle time), and decision templates.
• Every gate decision is supported by decision-quality artifacts: data readiness, evaluation plan/benchmarks, security posture, cost forecast, adoption plan.

C) Service catalog clarity (standardisation at scale)

• AI Lab services defined with SLAs, required artifacts, and volume forecasts (Sandbox, MLOps, Ethics, FinOps, AI CoE patterns).
• "Default route" for experimentation is the Toolkit/Sandbox-exceptions are explicit and time-boxed.

D) Value realisation discipline (BAU products)

• Quarterly value reviews for BAU products (Portfolio.AI, Marketing Translation) with funded optimisation backlog and measurable KPI movement.

E) Governance scaled (risk, ethics, and delivery discipline)

• Clear, enforceable principles distinguishing PoC vs pilot vs MVP vs GA; "done means" standards include wired release readiness, telemetry, and operational handover.
• Joint-success principles with vendors and business sponsors are explicit and measurable.

F) FinOps and run-cost control

• Cross-charging model maintained; GPU/compute strategy options presented early (trade-offs among performance, cost, and reliability).
• No "silent run-cost blowups": predictable budgets and proactive mitigations.

G) Singapore government co-funding success

• Maintain the co-funding deliverable map (milestones, KPI targets, eligible cost tracking).
• Produce structured quarterly reporting, demo days, and audit-ready evidence packs.
• Manage expectations and stakeholder engagement with government counterparts and ecosystem partners.

What you do

1) Portfolio strategy and prioritisation

• Define product strategy per initiative: problem statement, target users, adoption plan, KPIs, operating model.
• Run prioritisation with Sonal/Dan; explicitly manage trade-offs among value, feasibility, risk, time-to-market, and run-cost.

2) Governance and operating model

• Own Innovation Board mechanics: gate criteria, templates, decision logs, scorecards, and escalation paths.
• Define and enforce portfolio "definition-of-done" standards (including data/model contracts, security controls, observability, BAU handover).
• Ensure the Toolkit/Sandbox is not optional "nice-to-have," but the standard platform path.

3) Stakeholder leadership (business + architecture + security)

• Drive alignment and sign-offs with CDAO LT and Enterprise Architecture; ensure solution strategy is board-endorsed and operationally viable.
• Own narrative and executive-ready decision packs: trade-offs, cost, risk, adoption constraints, and recommended decision.

4) Vendor and partner shaping (without doing procurement's job)

• Translate roadmap into outcome-based requirements and success metrics that can be contracted.
• Ensure vendor delivery aligns with "industrialisation" expectations (security-by-design, documentation, monitoring, handover, cost controls).
• Define joint success measures and governance cadence with partners.

5) Adoption, ethics, and community flywheel

• Drive enterprise enablement motions that materially affect adoption: ethics hub, Copilot/GenAI usage guidance, community ideation intake, playbooks for responsible AI rollout.
• Ensure adoption is measurable (telemetry + feedback loops) and drives roadmap decisions.