What Can We Learn from Data & AI Talent in Kuwait’s Banks?

The table below is the starting point for the analysis. It compares the number of identified specialists with the number of Kuwait-based LinkedIn members associated with each bank. Across the nine banks, 278 publicly identified professionals equal 1.63% of the combined LinkedIn-associated membership used for this comparison.

These figures show what was publicly identified, not which bank has the largest or strongest Data & AI workforce. Search coverage differs between institutions, and the LinkedIn figures are not verified workforce populations. The table provides a structured view of the observed sample. Differences between banks may reflect search coverage as much as actual workforce differences, so they should not be interpreted as rankings.

Three banks shape much of the aggregate picture

NBK, KFH and Boubyan account for 182 identified professionals, or 65.5% of the sample. They also account for 51.2% of the 17,017 LinkedIn members used for comparison. Consequently, the overall role mix is particularly influenced by these three institutions.

The remaining six banks each have between 8 and 21 identified professionals. If those counts are close to their internal specialist staffing, continuity becomes a practical question: who covers a critical pipeline, model or reporting process when its main specialist is unavailable? Public-profile counts cannot tell us whether backup capacity exists.

Analytics and data science account for 101 people; BI, MIS and reporting account for another 68. Together, these categories make up 60.8% of the identified workforce. Engineering and architecture account for 27 people, or 9.7%.

This is a description of primary roles. It does not measure how people spend their time. An analyst may build pipelines; an engineering leader may spend time designing platforms. Each person is counted once, so that the total remains meaningful.

The useful question is how the analytical work is supported. Who maintains the data pipelines? Who resolves inconsistent definitions? Who takes responsibility when a model fails after deployment? How much of that work is performed by a shared team, a broader technology function or an external provider?

The dataset cannot establish an engineering shortage. It identifies a dependency worth examining: the work needed to make analysis repeatable and dependable may sit outside the roles most visible in this sample.

Leadership and governance raise different questions

The 46 leadership and strategy roles outnumber the 27 engineering roles. This does not mean that 46 managers supervise 27 engineers. These are broad role categories, and technical leaders are counted under leadership rather than again under their technical specialty. Senior roles may also be easier to discover. The practical question is whether leaders have access to the delivery capacity their mandates require.

Governance, data management and quality account for 20 people, or 7.2%. Team size alone does not show whether data ownership is clearly defined and consistently applied across the bank. Some responsibilities may belong to business owners and stewards whose main occupation falls outside this list. The next investigation should establish who owns definitions, who fixes quality problems at source, and whether those responsibilities receive enough time.

Thirteen AI/ML roles do not describe the full AI capability

Only 13 people are assigned to the dedicated AI/ML category. AI leaders, architects and analytical data scientists can contribute to AI while appearing elsewhere. A zero in this category therefore means no identified people assigned to that primary bucket; it does not establish that a bank has no AI activity.

Specialty describes what a person does. Placement describes where the role sits.

In this study, a specialist is someone whose main role focuses on Data & AI—not someone who simply uses data as part of a broader job.

Core

Part of a central or dedicated Data & AI team.Example: an analyst in an enterprise analytics function.

Embedded

A Data & AI specialist working inside a business or support function.Example: a BI specialist in Finance or an analyst in Risk.

Uncertain

The person qualifies for the study, but the available information does not show where the role is placed.

Placement is inferred from public information and does not represent verified reporting lines.

Using these definitions, 156 professionals (56.1%) are classified as core, 98 (35.3%) as embedded and 24 (8.6%) as uncertain. Just over a third of the identified specialists therefore work within business or support functions.

Two analysts may have similar technical skills, but someone in a central team and someone embedded in Risk may work with different priorities, stakeholders and access to shared support.

The patterns differ across banks. Boubyan has 39 of 50 identified people classified as core (78%). At KFH, 27 of 58 are core (46.6%) and 22 are embedded (37.9%), with 9 uncertain. These configurations raise different coordination questions, but do not tell us which organisation works better.

Data & AI capability extends beyond the identified workforce

Data owners, part-time stewards, champions and skilled business users can contribute without holding specialist roles. Banks may also rely on outsourced teams, technology vendors, consultants and other external delivery partners. These contributions should be recorded separately. Participation in a training programme is not equivalent to a full-time specialist, and specialist headcount alone is not a complete measure of enterprise capability.

This distinction also matters when returning to Redman: his embedded data managers cannot simply be equated with the embedded analysts in this dataset.

This analysis cannot determine how many Data & AI professionals each bank should employ. Redman’s framework helps explain why: staffing depends on the importance of data to strategy and the intensity of the data being managed. The percentages in his figure illustrate that relationship.

In the accompanying passage, Redman recommends 2% as a starting point for most organisations, with two-thirds to three-quarters of those roles embedded and the remainder in staff roles. The passage also suggests looking beyond total headcount to where data responsibilities sit, especially how much work is embedded in business teams.

However, comparing our 1.63% directly with his 2% would mix different role definitions and workforce populations. Our numerator is an incomplete sample spanning several specialist capabilities; our denominator is LinkedIn-associated membership. The difference between the percentages cannot be translated into missing employees.

Other benchmarks provide context

SYNQ’s study of 100 US and European technology scaleups reports a median data-team share of 3%, with most companies between 1% and 5%; fintech averages about 3.5%. It also discusses using engineering headcount as another denominator and recognising data work performed outside specialist roles. These are useful perspectives, but technology scaleups are not matched peers for Kuwait’s banks. [2]

Gartner’s June 2026 public abstract similarly presents ratios and size benchmarks as an initial reference and advises looking beyond them. The public abstract supports that principle; the full methodology is behind access controls. [3]

Neither source supplies a defensible target for the nine banks in this sample. A useful assessment would ask three things: how staffing compares with genuinely similar organisations, whether available teams can deliver and maintain the required work, and whether critical responsibilities have clear ownership and backup. If vendors or consultants contribute to Data & AI work, their contribution should be assessed from the work they actually perform—not simply from the existence or value of a contract or software subscription.

This exercise maps visible people and their assessed roles. A rigorous follow-up study should connect that map to the work each bank needs to perform.

The study should begin by validating the internal workforce with each institution, including the time and responsibilities of embedded specialists. It should then record attributable external support, distinguish it from software and infrastructure provision, and map the critical services, their owners and the skills required to keep them operating.

The outcome measures should follow the purpose of the work: more reliable reporting, shorter decision cycles, less recurring manual repair, better model performance, lower losses or attributable commercial gains. Counting deliveries alone will not establish their value.

Bank profitability is a much more distant measure. Differences in scale, business mix and other factors would need to be addressed before interpreting any relationship with staffing. An association would still not establish causation.

The question travels beyond banking

Other sectors can use the same approach: map specialist work, distinguish organisational placement from capability, identify contributions outside formal teams, and assess capacity against the decisions and services that matter. The banking percentages themselves should not become a template.

With institutional participation, a more rigorous and repeatable study could show where capability exists, where it depends on a few people and where work remains unsupported. That evidence could help direct investment more intelligently.

Redman’s passage raises a question about people. A follow-up study should establish whether those people have the time, support and authority to do the work their organisations depend on.

Snapshot: September 2026. The analysis covers 278 publicly identified professionals across nine Kuwaiti banks and applies the classifications described below. It is an exploratory public-data study, not an independently verified payroll census.