全职 Technical Lead — Enterprise AI Platform 工作, 薪水, Alliance Bank Malaysia Berhad Federal Territory 公司招聘中 - Ricebowl

Technical Lead — Enterprise AI Platform

Be the bank's technical owner of the Enterprise AI Platform as it is built. The platform is delivered with an implementation partner; you are the person on our side who shapes the design, challenges the decisions, and signs off that what was delivered is what we specified.

Technical Ownership of Platform Delivery

The vendor writes the code. You own whether what they build is right — day to day, in every design review and every integration decision, working within the platform architecture direction set by the AI Innovation Centre.

• Own the technical relationship with the delivery partner — review architecture and detailed designs, challenge decisions that weaken extensibility or increase lock-in, and hold delivery to the agreed standards rather than accepting what arrives.
• Translate architecture intent into buildable specification — turn platform design direction into integration patterns, API contracts and interface specifications precise enough for a vendor team to build against without repeated interpretation.
• Own acceptance — define what "done" means for each component, verify it against the specification, and refuse the handover where it does not hold. Do not accept a black box.
• Keep the bank capable of taking it over — ensure documentation, runbooks and knowledge transfer are sufficient for the bank to operate, extend and if necessary replace vendor-delivered components.

Integration & Solution Design

The platform only earns its keep if things plug into it cleanly — the productivity estate, core banking systems, and business-built AI assets alike.

• Own the integration design across the estate — define how the platform connects to the bank's productivity and collaboration environment, core systems and data sources, using REST, event-driven and protocol-based integration patterns.
• Own the seam with the productivity and low-code estate — design how agents and assistants built on the bank's Microsoft productivity and low-code platforms are registered, entitled and controlled through the Enterprise AI Platform rather than running unmanaged alongside it.
• Shape solution design for AI use cases — work with business and delivery teams to turn a defined use case into a technical approach that fits the platform, and steer it to the right build surface when it does not.
• Design for provider independence — ensure integration and control points hold when the underlying model, framework or cloud provider changes. A design that only works because of one vendor's native tooling is not one we can rely on.

Governance Controls & Assurance

Governance the bank can evidence to an examiner, built into the platform rather than asserted in a document.

• Convert AI governance policy into enforceable technical controls — work with Risk, Compliance and Security to specify the guardrails the platform must enforce: entitlement, logging, content safety, data residency, and cost and rate limits.
• Verify the controls actually work — test what the vendor implemented against what was specified, and produce the evidence that supports internal audit and regulatory examination.
• Keep the platform audit-ready as regulation evolves — track regulatory and policy change, assess the impact on platform design, and drive the required changes through delivery.

Research & Evaluation

A recurring responsibility rather than a full-time one — but the platform is only as good as the judgement behind what runs on it.

• Run structured evaluations — assess models, agent frameworks and retrieval approaches on measured accuracy, latency and cost, so platform decisions rest on evidence rather than vendor claims.
• Prototype to answer specific questions — stand up or commission a working proof of concept where a design question cannot be settled on paper, and be willing to discard it once answered.
• Track the field and translate it — follow developments in agent orchestration, retrieval, smaller and specialised models, and emerging integration protocols, and turn what matters into a concrete recommendation for the roadmap.
• Support technical evaluation in market scans and RFP processes — assess vendor and fintech solutions on architecture soundness, integration complexity and lock-in exposure, and defend that assessment to a procurement panel.

Technical Foundation

This is a technical role, but the day-to-day is design, review and specification rather than production engineering. The distinction that matters is between what you must be able to do yourself and what you must be able to judge.

Expected to do yourself

• Write specifications engineers can build from — API contracts, integration designs, sequence and architecture diagrams, and acceptance criteria that leave little room for interpretation.
• Work hands-on at prototype depth — call APIs, stand up a proof of concept, test an integration, and run an evaluation. Enough to verify a claim yourself instead of taking it on trust.
• Read and assess code and configuration — follow what the vendor has actually built well enough to tell whether it matches the design.

Expected to judge with confidence

• Cloud-native architecture and integration patterns — REST, event-driven and protocol-based integration including MCP and function-calling equivalents, applied well enough to sanity-check a design and identify where it will fail.
• AI and agent system design — retrieval-augmented generation, tool and function calling, agent orchestration, and where smaller or specialised models fit — enough to evaluate a vendor claim critically.
• The Microsoft productivity and low-code estate — how the bank's collaboration, identity and low-code platforms work, and how AI capability built on them must be governed and integrated. Prior depth is an advantage; the expectation is that you become conversant within your first two quarters.
• Regulated-industry data governance and DevSecOps — access control, lineage, retention and audit evidence, and how security and compliance controls are embedded in delivery pipelines, to the standard expected under BNM RMiT and PDPA.

Qualifications

Experience

• 5–8 years in solution architecture, platform engineering or technical delivery, with a genuinely hands-on background — someone who has built before moving into a design and oversight remit, not someone who has only ever reviewed.
• Experience as the client-side technical authority over a delivery vendor — holding an implementation partner to an architecture, rejecting work that did not meet specification, and being right about it.
• Demonstrated ownership of a platform or integration layer through to production in a regulated or enterprise environment, including responsibility after go-live.
• Practical exposure to building or designing LLM-based systems — beyond demonstration, with a view on what worked, what did not, and why.
• Regulated-industry experience strongly preferred — financial services or an equivalent environment where auditability and change control shape design decisions.

Skills & Attributes

• Willing to challenge, with evidence — pushes back on a vendor design, a benchmark or an internal assumption, and brings the analysis that supports the challenge.
• Independent technical judgement — works deep inside a vendor's ecosystem without assuming that vendor is the answer to every problem.
• Designs to be replaced — specifies components so a model, framework or provider can be swapped without rewriting the system around it.
• Strong written communication — produces design documents, specifications and decision records that engineers, vendors and governance bodies can all act on.
• Comfortable with ambiguity and pace — operates as technical owner in an environment where the problem is often defined while being solved.