Healthcare & MedTech
Clinical AI, remote monitoring, computer vision, wearable and IoT solutions, and regulated deployment.
— AI Strategy & Delivery Leadership
Build the AI capability. Focus on the right opportunities. Scale responsibly.
Most AI initiatives die between the demo and the deployment. I work the space in between — choosing the few use cases worth doing, standing up the Centre of Excellence to govern them, and staying until they run in production: measured, secured, adopted.
Technology leadership across AI strategy, CoE design and leadership, product innovation, and delivery. The work spans enterprise operating models, use-case portfolios, and AI systems in production.
01 AI Strategy
AI Strategy & Business Value
-
Identify material sources of value
Find the areas where AI or automation may add the most value.
-
Select the appropriate intervention
Match each opportunity with the approach that serves it best — AI, process design, or standard software.
-
Prioritise the portfolio
Weigh value, feasibility, data readiness, risk, complexity, and cost.
-
Establish the investment case
Include integration, infrastructure, people, oversight, and operations.
-
Target the right improvements
Target possible reductions in repetitive work, delays, errors, or cost-to-serve — assessed and measured in context.
-
Make it operable
Roadmaps, governance, investment priorities, operating models, and outcome measures.
02 Cross-Industry Consulting
Sector-fluent, tailored to every industry
Strategy, CoE design, governance, and delivery experience transfer across industries — and every solution is tailored to each sector's workflows, data, risks, regulation, and operating model.
Financial services & wealth management
AI operating models, advisor support, document and knowledge intelligence, investment-research synthesis, client insights, KYC intelligence, and decision support.
Telecommunications
Digital customer experience, chatbot and omnichannel intelligence, self-service adoption, and legacy-platform transformation.
Maritime & shipping
Predictive ship-management strategy, operational intelligence, equipment reliability, maintenance, safety, and information-heavy workflows.
SMEs
Small, well-scoped automation and AI opportunities that can be evaluated before committing to a larger program.
03 Featured — AI Centre of Excellence
An enterprise capability and operating model
An AI CoE connects strategy, people, platforms, and governance — so AI moves reliably from experiment to production.
How the practice helps
- Define the CoE's purpose, mandate, sponsorship, and alignment with business strategy.
- Design a federated model connecting shared AI expertise with business-unit ownership.
- Create processes for AI opportunity intake, assessment, prioritisation, governance, and portfolio review.
- Establish roles, talent requirements, skills frameworks, and AI literacy programs.
- Develop platform, data, architecture, integration, and reusable-service foundations.
- Create responsible AI, security, privacy, risk, and human-oversight practices.
- Operationalise model and prompt evaluation, MLOps/LLMOps, monitoring, and lifecycle governance.
- Track adoption and outcomes, and learn from pilots before scaling.
- Help teams progress from experimentation to controlled production use.
AI CoE
Enterprise capability & operating model
- Business strategy
- Use-case portfolio
- People and skills
- Data and platforms
- Governance and risk
- Production delivery
What the CoE builds, capability by capability — expand each area:
Data
Pipelines, quality and the layers beneath AI
Tools change — Fabric, Databricks, Snowflake, AWS — the layers stay the same.
Governance
Cross-cutting — access, lineage, cataloguing, certification, audit and compliance, applied to every layer below. Not one box at the bottom.
-
Source systems
SAP, CRM, on-prem databases, files, APIs, logs and IoT — structured or unstructured, reached by CDC, API or batch.
-
Ingestion
Batch or incremental loads; CDC or streaming; API or file-based — the pattern matched to the need.
-
Processing
Bronze → silver → gold: cleansing, conformance, joins, rules and quality checks — feature engineering for AI.
-
Storage
Lake, lakehouse or warehouse, with data marts for domains — on Fabric, Databricks, Snowflake, S3 or Redshift.
-
Semantic layer
KPIs, hierarchies and security — certified metrics in Power BI, Tableau or LookML; centralised or domain-owned.
-
BI consumption
Reports, dashboards, scorecards and embedded BI — decision support designed around the person using it.
-
AI consumption
Copilots, chat-with-data, forecasting, agents and anomaly detection — only as good as the data beneath.
Operations
Cross-cutting — monitoring, incidents, releases, cost, backup & DR, support. Architecture must think beyond build, into run.
Lakehouse
Flexible data foundation
Stores raw, curated data at scale — bronze, silver, gold. Structured, semi-structured or unstructured. Use when: high volume, mixed types, streaming, data-science access.
Decision cue: store flexibly
Warehouse / data mart
Structured analytics layer
Curated, dimensional models — facts, dimensions, aggregates. Optimised for SQL performance and repeatable reporting. Use when: financial, sales, ops or regulatory reporting.
Decision cue: structure carefully
Semantic layer
Business meaning layer
KPIs, hierarchies, definitions, row-level security, certification. What keeps every report calculating revenue the same way. Use when: self-service, Excel access, or AI needs trusted context.
Decision cue: define meaning centrally
AI Platform
Reusable foundations teams build on
A shared AI platform turns scattered projects into a portfolio: instead of every team rebuilding plumbing, they build on golden paths — paved roads from idea to production that have security, evaluation and monitoring built in.
Platform engineering
Common serving patterns for models and agents, integration and retrieval services, prompt and configuration management, and guardrails — built once, reused by every team.
MLOps / LLMOps
CI/CD for models and prompts, model and prompt registries, evaluation harnesses that gate every release, and monitoring for drift, quality, cost and latency once live.
Federation & standards
Standards and templates the CoE maintains; business-unit teams own their solutions. Self-service within guardrails, not central bottlenecks.
Build vs buy is weighed per layer and sized to the organisation's maturity rather than the vendor demo — the platform grows as the portfolio grows.
Security
Protecting AI systems and the data they touch
AI security is broader than IT security: the model, the prompts and the data around it are all part of the attack surface. The CoE sets the standards and checkpoints; delivery teams apply them in every project.
Threats in scope
Prompt injection and jailbreaks, data exfiltration through context and tools, model and prompt abuse, insecure agent tool-calls, and supply-chain risk in models and dependencies.
Controls
Data classification, access controls, secure model and prompt handling, input and output filtering, agent permission scoping, and red-team testing before release.
In the lifecycle
Security review built into evaluation and deployment gates — not bolted on after release — plus monitoring for new abuse patterns once systems are live.
Governance & Ethics
Intake to audit — with fairness and transparency built in
Governance
Intake and prioritisation gates, risk tiering proportionate to impact, clear ownership and escalation paths, human oversight where it matters, audit trails, and model/prompt evaluation governed through the lifecycle.
Ethics
Fairness and bias awareness, transparency and explainability, privacy by design, and honest communication about what a system can and cannot do — applied from use-case selection, not retrofitted.
Frameworks
Grounded in recognised frameworks — NIST AI RMF, ISO 42001, the EU AI Act — plus sector guidance: GDPR/PDPA, HIPAA, FDA SaMD, MAS TRM and FEAT. Familiarity, applied pragmatically.
The full responsible-AI view — frameworks, regulatory domains and ethics in the delivery lifecycle — is in Responsible AI below.
Infrastructure
GPU capacity — including in-house open-LLM inference
Where workloads should run is a portfolio decision, made per workload — not one answer for everything.
Hosting spectrum
External APIs, cloud-managed endpoints, or open-weight LLMs on your own GPU infrastructure — a private alternative to OpenAI or Claude when data cannot leave the tenancy, or sovereignty and residency apply.
GPU selection & sizing
Hardware and serving stack matched to the models, latency targets and volume — sized from measured workload, not guesswork, with the model families assessed for task fit, licence and footprint.
Operations & cost
Serving, monitoring, capacity and DR in your own environment — with total cost of ownership compared honestly against per-token API pricing over time.
The honest counterpoint: spiky or early-stage workloads often belong on external APIs first — the assessment weighs both against your volumes, risks and roadmap. Candidate model families include Qwen, Llama, Mistral, DeepSeek, Gemma and Phi — representative, not a fixed menu: selection is driven by the workload's task fit, language coverage, licence, footprint and evaluation results.
04 Innovation & Projects
AI, Wearable, and IoT innovation
Product thinking across hardware, sensors, data, AI, workflows, and deployment. Deployed work leads — prototypes, concepts, and patents follow, each labelled by its current stage.
-
Deployed
Computer-vision wound detection and tissue assessment.
-
Deployed
Real-time wearable and IoT telemetry from 300+ connected devices.
-
Deployed
Biomechanical motion analysis using wearable data.
-
Deployed
Population-health risk scoring from oxygen saturation, heart rate, HRV, respiratory rate, and sleep signals.
-
Deployed
AI applications integrated with clinical records and workflows.
-
Patented · No. 437540
A patented innovation in an absorbent multifunctional hygiene apparatus.
-
Prototype
Contactless vital-sign monitoring via camera-based remote photoplethysmography.
-
Prototype
Neurological signal processing using EEG data.
-
Concept
Battery-free, energy-harvesting SmartBandage sensing wound-fluid fullness, temperature, and pH, with patient alerts.
-
Concept
Smart-glass concepts for surgical recording and real-time tissue information.
05 Project Portfolio
Anonymised project portfolio
Organisations and clients are anonymised; every item is labelled, and figures are documented outcomes.
Clinical AI and connected care
Deployed workFive clinical AI platforms Deployed
- Focus area
- Cardiovascular, neurological, orthopedic, wound-care, and remote-monitoring pathways.
- Type of work
- Production platform delivery and portfolio governance.
- AI / technology approach
- Clinical predictive models, computer vision, wearable and IoT telemetry, regulated delivery.
- Supported outcome
- Regulatory-approved production across three hospital networks, zero regulatory incidents — documented outcome.
Parkinson's dyskinesia event detection Deployed
- Focus area
- Objective detection of dyskinesia events.
- Type of work
- Clinical AI development and validation.
- AI / technology approach
- Machine-learning event detection on neurological signal data.
- Supported outcome
- Validated in a 60-patient clinical trial.
Orthopedic biomechanical motion analysis Deployed
- Focus area
- Assessing movement in orthopedic care.
- Type of work
- Deployed clinical AI capability.
- AI / technology approach
- Motion analysis from wearable sensor data.
- Supported outcome
- Used in deployed orthopedic pathways.
Population-health risk scoring Deployed
- Focus area
- Risk signals across a monitored population.
- Type of work
- Deployed predictive analytics.
- AI / technology approach
- Risk models from vital and sleep signals.
- Supported outcome
- Deployed for population-health monitoring.
Remote patient monitoring platform Deployed
- Focus area
- Monitoring patients outside hospital settings.
- Type of work
- Deployed connected-care platform.
- AI / technology approach
- Real-time telemetry from 300+ wearable and IoT devices.
- Supported outcome
- 45% faster healing outcomes and reduced bed occupancy — documented outcome.
Wound-care and medical-device innovation
Deployed workComputer-vision wound assessment Deployed
- Focus area
- Consistent, objective wound assessment.
- Type of work
- Deployed clinical AI capability.
- AI / technology approach
- Wound detection and tissue classification.
- Supported outcome
- The platform holds FDA Class I and HSA Class A registration, as stated in the CV.
- Live product
- woundaid.co ↗
Contactless vital-sign monitoring Prototype
- Focus area
- Vital signs without physical contact.
- Type of work
- Prototype exploration.
- AI / technology approach
- Camera-based remote photoplethysmography.
- Supported outcome
- Prototype built to explore real-world performance.
Battery-free SmartBandage concept Concept
- Focus area
- Monitoring wounds between clinical visits.
- Type of work
- Concept design.
- AI / technology approach
- Energy-harvesting sensing of wound-fluid fullness, temperature, and pH.
- Supported outcome
- Concept designed for future development.
Smart-glass surgical concepts Concept
- Focus area
- Capturing and augmenting information during surgery.
- Type of work
- Concept design.
- AI / technology approach
- Surgical recording and real-time tissue information.
- Supported outcome
- Concepts designed for future development.
EEG signal-processing integration Prototype
- Focus area
- Interpreting neurological signals in context.
- Type of work
- Prototype exploration.
- AI / technology approach
- EEG signal processing integrated with clinical workflows.
- Supported outcome
- Prototype built to explore real-world performance.
Clinical workflow co-design Deployed activity
- Focus area
- Making AI usable in clinical routines.
- Type of work
- Deployment support.
- AI / technology approach
- Workflow co-design and AI adoption workshops.
- Supported outcome
- Delivered alongside clinical AI platforms.
Financial services
Strategy & operating-model designStrategy, operating-model, and architecture design for enterprise AI delivery.
Enterprise AI strategy and CoE operating-model design Strategy / Blueprint
- Focus area
- AI capability in a regulated banking environment.
- Type of work
- Strategy and operating-model design.
- AI / technology approach
- AI strategy and CoE operating-model blueprint.
- Supported outcome
- A blueprint for capability, governance, and delivery foundations.
AI portfolio planning Strategy / Blueprint
- Focus area
- Deciding where AI creates value.
- Type of work
- Strategy and portfolio design.
- AI / technology approach
- Portfolio across advisor enablement, research synthesis, client insights, document intelligence, KYC intelligence, and decision support.
- Supported outcome
- A prioritised portfolio for phased delivery.
Data-readiness planning Strategy / Blueprint
- Focus area
- Data prepared to support AI at scale.
- Type of work
- Architecture and readiness planning.
- AI / technology approach
- Quality, access, lineage, governance, semantic understanding, evaluation, and pipelines.
- Supported outcome
- A data-readiness plan for enterprise AI.
AI industrialization blueprint Strategy / Blueprint
- Focus area
- Moving models to controlled production use.
- Type of work
- Architecture blueprint.
- AI / technology approach
- Reusable services, MLOps/LLMOps, evaluation, monitoring, security, human oversight.
- Supported outcome
- An industrialization blueprint for delivery.
Maritime and shipping
Strategy & architecture blueprintA strategy and architecture blueprint, charting the path toward future deployment.
Predictive ship-management transformation blueprint Strategy / Blueprint
- Focus area
- Reliability, safety, and operations across a fleet.
- Type of work
- Strategy and architecture blueprint.
- AI / technology approach
- Blueprint for 180+ vessels: operational intelligence, document simplification, port monitoring, reliability, predictive maintenance.
- Supported outcome
- An opportunity map and blueprint for phased transformation.
Document intelligence and knowledge architecture concepts Blueprint · Concept
- Focus area
- Document-driven operational workflows.
- Type of work
- Concept design within a blueprint.
- AI / technology approach
- Document intelligence, RAG, semantic search, knowledge graphs, agent orchestration.
- Supported outcome
- Concepts defined for evaluation and future deployment.
Predictive decision-support concepts Blueprint · Concept
- Focus area
- Timely operational decisions across voyages.
- Type of work
- Concept design within a blueprint.
- AI / technology approach
- Decision support combining operational, vessel, voyage, weather, telemetry, and maintenance data.
- Supported outcome
- Concepts defined for evaluation and future deployment.
Telecommunications and digital transformation
Delivered workDigital customer-service chatbot and omnichannel intelligence Deployed
- Focus area
- Rising service demand across digital channels.
- Type of work
- Delivery initiative.
- AI / technology approach
- Chatbot and omnichannel intelligence.
- Supported outcome
- 25% improvement in digital self-service adoption and reduced cost-to-serve, as stated in the CV — for this documented initiative.
Large-scale digital transformation Deployed
- Focus area
- Bringing legacy platforms into a modern digital experience.
- Type of work
- Transformation delivery.
- AI / technology approach
- Legacy modernisation, data migration, digital platforms, omnichannel experiences.
- Supported outcome
- Delivered at scale across the documented program.
06 AI opportunities for growth businesses
AI opportunities for growth businesses
Other engagements — each with a sensible first step and the considerations that set it up for success.
Other engagement
Invoice and document handling
Opportunity: free up staff time from manual extraction and sorting. First step: pilot one document type — extract key fields for staff review. Consider: formats, privacy, human review, integration.
Other engagement
Internal knowledge assistant
Opportunity: help staff find answers in policies and manuals quickly. First step: an assistant over a few approved documents, with links to sources. Consider: content currency, permissions, cited answers.
Other engagement
Customer enquiry support
Opportunity: respond to enquiries faster with prepared drafts. First step: categorise enquiries and draft responses for a person to review and send. Consider: accuracy, customer data, human ownership.
Other engagement
Lead follow-up prioritisation
Opportunity: give every enquiry a timely response. First step: organise enquiries by agreed criteria and flag those needing timely response. Consider: clear criteria, with people deciding the follow-ups.
Other engagement
Stock and demand alerts
Opportunity: keep stock levels well matched to demand. First step: alerts for unusual demand or low-stock risk from existing sales and inventory data. Consider: data quality, seasonality, staff review before reordering.
Other engagement
Scheduling support
Opportunity: build work-order and field-service schedules in minutes. First step: propose schedules from availability, location, and business rules for a person to approve. Consider: real rules, edge cases, acceptance.
Other engagement
Quality inspection support
Opportunity: make visual inspection faster and more consistent. First step: explore computer vision for one defined inspection task with human review and testing first. Consider: image quality, variation, false positives.
Other engagement
Meeting and service-record summaries
Opportunity: spend meeting time on decisions while the summary drafts itself. First step: turn approved recordings or notes into draft summaries and action lists. Consider: privacy and consent, human review, retention policy.
Other engagement
AI readiness and workflow review
Opportunity: gain clarity on where AI delivers the most value. First step: a short workflow review that identifies where process redesign or existing software offers the best path. Consider: cost, capability, change effort.
Each engagement is a well-scoped first step, evaluated on its own merits — building confidence before committing to a larger program.
07 Delivery Approach
How the work runs
-
Discover
Understand goals, workflows, users, cost pressures, data, and constraints.
-
Prioritise
Assess value, feasibility, readiness, cost, complexity, and risk.
-
Organise
Design the CoE, operating model, governance, people, platforms, ownership.
-
Design and build
Architect, prototype, or guide implementation of selected solutions.
-
Measure and improve
Evaluate adoption and outcomes, then refine or scale responsibly.
08 Responsible AI
Responsible AI and production readiness
Responsible AI runs across the lifecycle — governance, privacy, security, human oversight, explainability, auditability, model and prompt evaluation, monitoring, and regulatory readiness. Oversight designed in from the start and monitored after release — AI that is controlled, auditable, and ready to stay in production.
Governance designed for Board assurance, regulatory scrutiny and production accountability.
Frameworks and regulatory domains reflected in professional experience include those listed here — areas of working familiarity gained through hands-on practice.
AI Security
Protecting AI systems and the data they touch — secure model and prompt handling, data protection, access controls, adversarial input awareness such as prompt injection, and security review built into evaluation and deployment.
AI Governance
Clear ownership, policies, and checkpoints — intake and prioritisation gates, risk assessment, human oversight and escalation paths, audit trails, and monitoring that keeps models accountable in production.
AI Ethics
Fairness, transparency, and respect for the people affected — bias awareness, explainability, privacy by design, and honest communication about what a system can and cannot do.
09 About me
Strategy, capability and delivery — under one leadership mandate
Senior technology and transformation executive with more than 22 years of experience translating business strategy into enterprise-scale digital, technology, data, cloud and AI transformation programmes across healthcare, telecommunications and regulated sectors throughout APAC.
A proven record of leading complex, multi-workstream programmes; building and scaling high-performing teams; managing multimillion-dollar budgets and strategic partnerships; and aligning C-suite, Board, business, technology, clinical, regulatory and delivery stakeholders.
Led a multimillion-dollar enterprise transformation across four geographies through a multi-partner delivery model, securing significant Board investment through a structured investment strategy and delivering five production platforms. Brings strategic judgement, commercial discipline, programme governance, digital innovation, operational improvement and hands-on technology leadership to deliver sustainable, measurable outcomes.
Global delivery experience spans APAC, Europe and North America, leading programmes and platform teams across time zones. Current doctoral research in AI and machine learning focuses on enterprise agentic AI architecture, sovereign AI frameworks and ethical AI for regulated industries.
- AI strategy and enterprise transformation
- AI Centre of Excellence design and leadership
- Cross-industry AI portfolio planning
- AI, Wearable, IoT and product innovation
- Production engineering and responsible deployment
10 Contact
Discuss your AI mandate
Whether establishing an AI Centre of Excellence, prioritising an investment portfolio, or assessing a strategic use case, arrange a confidential initial discussion. Engagements run across regions and time zones, remotely and on site where required.