Home / AI & digital transformation

I've shipped the AI I now help others govern.

Nine years at GCB Bank, from business process analyst to data scientist to automation engineering lead. Every project below went to production in a regulated bank.

GCB Bank · Automation Engineering Lead · Nov 2025

The bank's first enterprise GenAI assistant

Frontline staff were losing time hunting through SOPs and waiting on supervisors for answers. I designed, built and deployed an assistant that answers from the bank's own procedures.

−90%

information-retrieval time

Read the case study

Context

3,500+ customer-facing staff, hundreds of SOPs, and supervisors acting as the search engine.

What I owned

Design, build, deployment and the risk-tiered roadmap.

Approach

  • Start with retrieval over approved SOPs only — low risk, high value
  • Defer high-risk agentic features until controls exist
  • Deliver against a fixed date

Execution

Launched in November 2025, on schedule. The phased, risk-tiered roadmap was adopted as the CoE standard for later AI projects.

Results

3,500+staff with access
−90%time to find information
−95%dependence on supervisors
On timeNov 2025 launch

GCB Bank · 2025

A Centre of Excellence for intelligent automation

Eight business units were each asking for automation separately. I set up one operating model to take requests in, assess them, gate them and deliver them.

8

business units on one pipeline

Read the case study

What I owned

The operating model: intake and assessment framework, Risk–Compliance–Internal Control gates, delivery pipeline. I chaired cross-unit prioritisation and led the engineering team.

Approach

Agile sprint planning across units; governance gates as a pipeline stage rather than an afterthought.

Results

8business units coordinated
3control functions gating every build

GCB Bank · 2025

Failed bank-to-wallet transfers: refunds in 3 hours, not 24

When mobile-money transfers failed, customers waited a day for their money back while staff reconciled by hand.

24→3h

customer reimbursement time

Read the case study

Execution

Built an RPA solution that reconciles and reverses failed bank-to-wallet transactions automatically.

Related builds

  • The bank's first fully digital onboarding and loan-origination flow (process design and architecture), opening the diaspora market
  • An automated credit-risk monitoring and escalation system

Results

−87.5%reimbursement time (24h → 3h)
1stfully digital onboarding at the bank

GCB Bank · Data Scientist · 2024–2025

Predictive models that moved the P&L

Customer lifetime value and churn models, built with Risk, Compliance and IT so they could go to production in a bank.

25.3%

FY2024 profit growth the models contributed to

Read the case study

What I owned

Model development and model governance with Risk, Compliance and IT. Advised the Deputy Managing Director on enterprise AI/ML strategy.

Execution

CLV optimisation and churn reduction models, plus ML-driven dynamic pricing recommendations that fed retention and pricing decisions.

Results

3models in production
−34%churn
25.3%FY2024 profit growth (contributed)

GCB Bank · Business Process Analyst · 2022–2024

Process before automation

Before anything gets automated, it has to be understood. I built the framework that decided what to automate, and fixed the processes that weren't ready.

−35%

processing errors

Read the case study

Execution

  • RPA assessment framework scoring feasibility and ROI
  • SOPs for 10+ critical departments, including IT Ops, Risk and Compliance
  • Process-improvement curriculum delivered to 120+ staff

Earlier

As Assistant Maintenance Specialist (2017–2022): co-authored the maintenance policy and redesigned vendor management, cutting opex by an average of 32% a year.

Results

+85%team productivity across 7 assessments
−35%account/loan errors
−15%cycle time
120+staff trained

Planning an AI rollout?

I'll help you sequence it so the first release delivers value and the risky parts wait for their controls.

Book a callTier a use case