Program 01 · Domain specialisation

SCIKIQ Certified Data and AI EngineerBanking & Financial Services

Banks do not have an AI problem; they have a deployment problem. The models work. What breaks is core-system access, model risk sign-off, lineage for a regulator and a cost per case nobody measured. This specialisation takes you through the banking value chain, the architecture that actually gets approved, and the use cases that move a P&L line.

80 hrstaught
40hands-on AI labs
8assessment tests
+20 hrs2 E2E capstones

Talk to us about a cohort See the assessments

$200-340bnAnnual gen-AI value potential in bankingMcKinsey

60-70%Of a bank’s cost base sitting in end-to-end operationsMcKinsey

40-60%Reduction in AML false positives reported by agentic screening

3-15%Higher revenue per RM where a frontline domain is rewiredMcKinsey

Part 1 · The business

The value chain, the people, the trends

Before you design an agent, you need to know which decision you are automating, who owns it today, and what it costs when it is wrong. Part 1 maps the banking value chain end to end, the people accountable at each stage, and what the strategy houses are seeing in 2026.

The banking value chain

Six stages, each with its own decisions, failure modes and owners. Select a stage.

Who you are actually building for

A SCDAI engineer deploys into someone’s week. These are the five people whose week changes. Select one to light the stages they own.

What the strategy houses are seeing

The 2026 read from McKinsey, BCG, Bain and HBR — functional and firm-wide.

McKinsey60-70% of cost base

Operations is the value pool, not the demo

End-to-end operations represent an estimated 60-70% of a bank’s cost base, which is why reusable “operations transformer” agents beat one-off pilots.

McKinsey3-15% revenue per RM

Frontline rewiring pays in months

Banks that rewire a single frontline domain — not the whole bank — report materially higher revenue per relationship manager and lower cost to serve.

McKinsey9-15% of operating profits

The headline number is productivity

Gen AI could add $200-340bn annually in banking, equivalent to 9-15% of operating profits, overwhelmingly from productivity rather than new revenue.

BCGProcess-first, not tool-first

Redesign the process, do not decorate it

The firms capturing returns redesign core processes around AI capability. Bolting an assistant onto a process built for humans reliably underdelivers.

BainFewer, deeper bets

Concentrate, do not sprinkle

Value accrues where AI is embedded behind the few tactics that matter, rather than spread thinly across a portfolio of pilots nobody scales.

HBRExplainability > accuracy

Trust is the deployment constraint

In regulated decisions, adoption tracks explainability and recourse far more than raw accuracy. The reviewable agent beats the marginally better one.

Where these numbers come from — 4 sources

Consulted when this program was written, in July 2026. Each entry names the claim on this page that it backs, so the pairing stays checkable if the copy is later edited.

McKinseyCapturing the full value of generative AI in banking$200-340bn annual value, 9-15% of operating profits
McKinseyAI in Asia: Reimagining banking operations through agentic AIEnd-to-end operations as 60-70% of the cost base
McKinseyAgentic AI is here: is your bank’s frontline team ready?3-15% higher revenue per relationship manager
IndustryAgentic AI drives next phase of AML innovation40-60% reduction in AML false positives

Part 2 · Data and meaning

Domains, ontology, taxonomy and governed terms

Part 2 is the half most training skips. A banking agent lives or dies on which systems it can reach, which data it is allowed to see, and whether every action it takes can be evidenced two years later.

The data landscape you will actually meet

Six source families, each with its own access path, latency and political owner.

Core banking

Accounts, balances, postings. Batch-oriented, jealously guarded, rarely queryable in real time.

Finacle, Temenos T24, FIS, Mambu

Payments & messaging

ISO 20022 and SWIFT MT flows, repair queues and settlement state.

ISO 20022, SWIFT, SEPA, RTP

Customer & CRM

Relationship, KYC profile, interactions. The golden-record problem in its purest form.

Salesforce FSC, MDM hubs

Risk & finance marts

Exposures, provisions, capital. Governed, reconciled and slow to change.

Data warehouse, BCBS 239 marts

Unstructured corpus

Policies, credit files, contracts, memos, regulatory text. Where most of the answer lives.

SharePoint, ECM, loan files

Surveillance & case

Alerts, cases, investigations, SAR history. The system of record for “what did we do”.

Actimize, case management

The semantic layer

A banking agent that cannot tell a party from a counterparty, or an exposure from a balance, will produce confident nonsense. These are the domains, the graph and the definitions the agent must be grounded on before it is allowed to answer anything.

Data domains

Six subject areas, each with the entities it holds, its critical data elements and the function accountable for it.

Party

Every person and legal entity the bank deals with, and the roles they play.

CustomerCounterpartyBeneficial ownerRelated party

Critical data elements: Legal entity identifier; Tax identifier; Ownership percentage
Owned by: Customer data / KYC

Arrangement

The contracts the bank holds with parties: accounts, loans, facilities, limits.

AccountLoanFacilityLimitCollateral

Critical data elements: Account identifier; Committed amount; Maturity date
Owned by: Product control

Transaction

Movements of value and the messages that carry them.

PaymentPostingTradeSettlement

Critical data elements: Transaction reference; Value date; Amount and currency
Owned by: Operations

Risk & exposure

What the bank stands to lose, by counterparty, product and scenario.

ExposureRatingProvisionCapital charge

Critical data elements: Probability of default; Exposure at default; Loss given default
Owned by: Risk

Financial crime

Alerts, cases, investigations and the regulatory filings they produce.

AlertCaseSuspicious activity report

Critical data elements: Alert disposition; Typology code; Filing date
Owned by: Financial crime compliance

Product & policy

What the bank sells and the rules governing how it is sold and serviced.

ProductPolicy documentPricing schedule

Critical data elements: Product code; Policy version; Effective date
Owned by: Product & compliance

Ontology & knowledge graph

The entities and the typed relationships between them — what the agent traverses instead of guessing joins. Select any entity.

Core entityEventReference / classification

Taxonomies

The classification hierarchies that make records comparable across systems.

Product hierarchy

Line of business
Product family
Product
Variant

Retail → Deposits → Savings account → Notice saver

Customer risk rating

Risk band
Driver
Factor

High → Jurisdiction → Sanctioned country exposure

Financial crime typology

Category
Typology
Indicator

Money laundering → Layering → Rapid movement through multiple accounts

Governed business terms

The definitions an agent must use rather than invent. Most wrong answers in this industry are a term used loosely.

Customer: A party with an active servicing relationship. Not every party is a customer, and prospects are not.

Counterparty: A party the bank has exposure to, which may or may not be a customer.

Exposure: Amount at risk to a counterparty at a point in time — not the same as outstanding balance.

Outstanding balance: Drawn amount on an arrangement now. Excludes undrawn commitment.

Committed amount: The full limit the bank is obliged to make available, drawn or not.

Beneficial owner: The natural person who ultimately owns or controls a legal entity.

Alert disposition: The outcome of reviewing a monitoring alert: closed, escalated or filed.

Default: The regulatory event triggering loss recognition — defined by policy, not by intuition.

Part 2 · The architecture

How it is actually built

The layers a deployment needs, what each one holds, and what the engineer owns there.

The reference architecture

Seven layers. Click one to see what it holds and what the engineer owns there.

Typical tech stack

Models

Frontier API models under enterprise agreements
Smaller models for extraction and classification
Private endpoints, no training on your data

Grounding

Vector store with hybrid search
Knowledge graph for relationships
Reranking for precision

Agents

LangGraph / framework of record
MCP servers over core systems
Human approval gates

Data

CDC into a governed lakehouse
Catalogue, lineage and glossary
Governed metric layer

Assurance

Eval harness with labelled sets
OpenTelemetry GenAI tracing
Model risk documentation

Runtime

Private cloud or bank tenancy
Least-privilege service identity
Full request/response retention

Constraints that shape the design

Model risk management

SR 11-7 style validation applies. If you cannot document intended use, limitations and monitoring, it does not ship.

Data residency & secrecy

Bank secrecy and residency rules often forbid the obvious architecture. Design for in-region inference from day one.

Explainability & recourse

Adverse decisions need a reason a customer can be given and a human can defend.

Evidence retention

Assume every prompt, retrieval and action must be reproducible years later. Retention is an architecture decision, not a log setting.

Part 3 · Getting it done

Where the business and the technology meet

Part 3 is where Parts 1 and 2 meet. Every use case below names the value-chain stage it changes and the architecture layer that carries the weight — then you build it, evaluate it and hand it over.

Use cases, mapped to the chain and the architecture

Filter by stage or by how much autonomy the use case has earned. Selecting a use case jumps the value chain and the architecture to the stage and layer it depends on.

Value chain stage

Autonomy earned

12 use cases

The delivery playbook

How a Senior SCDAI Engineer actually runs a banking deployment, week by week.

Week 1 — find the decision

Sit with the queue. Count the volume, the handle time and the rework. Pick the one decision where volume × error cost is highest and access is achievable.

Week 2 — thin slice on their data

Build end to end on a real, messy sample under an NDA sandbox. No synthetic demo. The goal is to discover the data truth early, not to impress.

Weeks 3-5 — ground and evaluate

Build the retrieval layer properly, label an eval set with the business owner, and publish a baseline the team agrees is fair.

Weeks 6-8 — earn autonomy

Start at assist. Move to approve. Only automate the band where evals and human review agree, and write the model risk pack as you go.

Weeks 9-12 — hand over

Runbook, rollback, on-call, eval suite in CI and a named owner. Success is that they change it next quarter without calling you.

The curriculum

8 modules, 80 taught hours, 40 hands-on labs and 8 assessments — every lab provisioned and graded by the SCIKIQ Agentic AI Playground. Open a module to see its labs.

01The SCDAI delivery model in a regulated bankHow a certified data and AI engineer works inside an institution where every change has an owner and an auditor.10 hrs · 5 labs

The Delivery mandateDiscovery in a regulated queueSizing volume × error costAccess negotiationThin-slice scopingStakeholder map

Shadow a queueInstrument a simulated servicing queue and quantify volume, handle time and rework.

Decision inventoryMap one banking workflow into discrete decisions and pick the automatable one.

Access path memoDocument how you would reach core, CRM and the document store, with latency budgets.

Thin-slice scopeWrite a one-page scope for a two-week slice with an explicit non-goal list.

Stakeholder mapIdentify the approver, the blocker and the champion for a named use case.

Assessment 01 — Operating model & discovery25 items · 35 min · pass 70%

02Banking data foundationsCore, payments, CRM, risk marts and the unstructured corpus — what each will and will not give you.10 hrs · 5 labs

Core banking access patternsISO 20022 & payments dataCustomer golden recordBCBS 239 lineageDocument corporaEntitlements

Core extractionPull account and posting data through an API stub and model its latency limits.

ISO 20022 parsingParse pain/pacs messages and normalise them into a queryable structure.

Entity resolutionResolve a customer across three systems that disagree, and score the matches.

Lineage captureInstrument a pipeline so every field can be traced back to source.

Entitlement modelDesign row and column entitlements for an agent acting on behalf of a user.

Assessment 02 — Banking data landscape25 items · 35 min · pass 70%

03Context engineering for financial languageFinancial text is adversarial: dense, templated, full of numbers that must not be paraphrased.10 hrs · 5 labs

Prompt and context designNumeric fidelityStructured output contractsPolicy interpretationFew-shot from real filesRefusal behaviour

Structured extractionForce schema-valid JSON out of a messy scanned credit file.

Numeric fidelityBuild a check that fails the run when a figure is not traceable to source.

Policy promptGround a policy question so the answer quotes the clause, not the model.

Refusal designMake the agent decline out-of-policy questions instead of improvising.

Context budgetCut context by half with no accuracy loss on the eval set.

Assessment 03 — Context engineering25 items · 35 min · pass 70%

04Grounding: RAG and knowledge graphs in bankingPolicy, credit files and relationships — three different retrieval problems in one deployment.10 hrs · 5 labs

Chunking policy documentsHybrid searchRerankingGraphRAG for exposureCitation contractsRetrieval evaluation

Policy corpusChunk and index a policy set, then measure retrieval recall on labelled questions.

Hybrid + rerankAdd BM25 and a reranker; quantify the precision gain.

Exposure graphTraverse counterparty relationships to answer a question vector search cannot.

Citation contractGuarantee every claim carries a resolvable source reference.

Retrieval evalsBuild a labelled retrieval set and gate builds on recall@k.

Assessment 04 — Retrieval & grounding25 items · 40 min · pass 70%

05Agent design for regulated workflowsWhere autonomy is appropriate, where approval is mandatory, and how to make the difference explicit.10 hrs · 5 labs

Workflow vs agentTool designApproval gatesState and memoryMulti-agent patternsFailure containment

Alert triage agentBuild a triage agent that ranks alerts and states its reasons.

Approval gateInsert a human approval step that blocks the action, not just the message.

Idempotent toolsMake an action tool safe to retry and prove it under duplicate calls.

Supervisor patternSplit extraction, analysis and drafting across coordinated agents.

ContainmentAdd budget, loop and blast-radius limits that stop a runaway run.

Assessment 05 — Agent design & autonomy25 items · 40 min · pass 70%

06Integrating with the bankMCP servers over core systems, service identity, and the operational reality of a change window.10 hrs · 5 labs

MCP server designLeast-privilege identityRate limits and batch windowsCase management write-backError handlingChange management

MCP over coreExpose a read-only core banking tool over MCP with typed schemas.

Write-back toolWrite an investigation note to a case system with full audit metadata.

Least privilegeScope a service identity so the agent cannot exceed the user behind it.

BackpressureHandle rate limits and batch windows without losing work.

Failure modesDesign retries and dead-letter handling for a flaky downstream core.

Assessment 06 — Systems integration25 items · 35 min · pass 70%

07Model risk, fairness and controlsThe module that decides whether your build ever reaches production.10 hrs · 5 labs

Model risk documentationFair lending and disparate impactOWASP LLM Top 10Prompt injection in document pipelinesAdverse action reasoningChallenger review

Model risk packProduce intended use, limitations and monitoring documentation for your agent.

Bias probeTest a decision agent for disparate outcomes across protected proxies.

Injection defenceDefend an extraction pipeline against instructions hidden in a submitted PDF.

Adverse actionGenerate a decline reason that is accurate, specific and defensible.

Red teamRun an OWASP LLM Top 10 pass and fix the two worst findings.

Assessment 07 — Risk, fairness & security30 items · 45 min · pass 70%

08Production, observability and costDeployment, evidence, drift and the unit economics that decide whether year two happens.10 hrs · 5 labs

Eval suites in CIOTel GenAI tracingDrift and judge calibrationCost per caseRollbackHandover

Eval gateBlock a deploy on an eval regression and prove the gate works.

Trace the runInstrument an agent with OTel GenAI conventions and debug from spans.

Judge calibrationCompare LLM-judge scores to human labels and correct the drift.

Cost per caseMeasure and then halve the cost of one completed case.

Handover packWrite the runbook, rollback path and owner list for a live agent.

Assessment 08 — Production & operations30 items · 45 min · pass 70%

Two capstones — 20 hours

End-to-end agent design, development, deployment and testing, 10 hours each. Reviewed by a Senior SCDAI Engineer against a published rubric.

Capstone A — Financial crime investigation agent

Build an agent that triages an AML alert, assembles evidence across four systems, drafts a narrative with citations and routes it for investigator sign-off. Prove recall on known true positives.

Capstone B — Credit origination copilot

Build an agent that spreads statements, extracts covenants, checks policy and drafts a credit memo with page-level sources — then evidence it against a model risk checklist.

What you deliver, and how it is marked

A running agent, deployed on the program's target stack, reachable by a reviewer.
An architecture note: boundaries, tools, autonomy level and the trade-offs you took.
A versioned eval set with a baseline and your final scores.
A trace walkthrough of one success and one deliberate failure.
Cost per task, measured — not estimated.
A runbook and rollback path a client team could operate on Monday.

Problem fit15%Is this a decision worth automating, scoped to a defensible boundary?

Architecture20%Right pattern for the constraints; autonomy earned, not assumed.

Grounding & accuracy20%Measured groundedness against the eval set, with the failures named.

Safety & compliance20%Domain controls actually enforced, and demonstrably so.

Operability15%Traces, runbook, rollback, cost — could someone else run this?

Communication10%Can a client executive follow the reasoning in ten minutes?

One lab, start to finish

Every lab in this program follows the shape below. This is lab 04 in full — the brief you are given, the environment that is provisioned for you, the code you start from and the assertions that decide whether you passed.

Module 04 · Grounding: RAG and knowledge graphs in banking

Exposure graph traversal

Answer a question vector search cannot: which customers are exposed to a delayed counterparty.

The brief

You are given a seeded graph of parties, ownership links, facilities and collateral, plus a policy corpus. A counterparty has just been downgraded. Return every customer with direct or indirect exposure to it, the path that creates the exposure, and the aggregate amount — with each hop citable.

Provisioned for you

Neo4j with 12k seeded party and facility nodes
Vector index over the policy corpus
Model endpoint pre-wired
Held-out question set (40 labelled)

You start from this

def exposed_customers(graph, counterparty_id, max_hops=3):
    """Return [(customer_id, path, amount)] for every exposure route."""
    # 1. traverse ownership and guarantee edges outward from the counterparty
    # 2. collect facilities whose obligor sits on any returned path
    # 3. aggregate committed - drawn, and keep the path for citation
    raise NotImplementedError

What the grader asserts

Finds all 7 indirect exposures in the held-out set, not only the 3 direct ones
Every returned exposure carries a full node path, not just an endpoint
Aggregate amount uses committed, not drawn, and matches the reference within 0.5%
Terminates within max_hops and does not revisit nodes on cyclic ownership
Returns empty with a stated reason when the counterparty is unknown, rather than guessing

Pass barAll five assertions green, and recall on the 40-question held-out set at or above 0.9.

StretchExpose the traversal as an MCP tool with a typed schema, and have an agent answer the question in natural language with the path cited.

Assessment

Assessed, not attended

Every module ends with a timed, randomised assessment delivered through the SCIKIQ Agentic AI Playground. The certificate requires a pass on all of them plus two reviewed capstones.

Assessment blueprint

What each test covers, how long it runs, and how many items are currently in the versioned bank behind it.

#	Assessment & coverage	Items	Time	In bank
01	Operating model & discoveryDelivery mandate and boundariesDiscovery techniqueValue sizingScoping and non-goals	25	35 min	4
02	Banking data landscapeCore and payments dataEntity resolutionLineage and BCBS 239Entitlements	25	35 min	4
03	Context engineeringContext designStructured outputsNumeric fidelityRefusal and scope	25	35 min	4
04	Retrieval & groundingChunking and indexingHybrid retrievalGraphRAGRetrieval metrics	25	40 min	4
05	Agent design & autonomyWorkflow vs agentTool and approval designMulti-agent patternsContainment	25	40 min	4
06	Systems integrationMCP tool designIdentity and least privilegeResilienceWrite-back safety	25	35 min	4
07	Risk, fairness & securityModel risk managementFairness testingLLM securityExplainability	30	45 min	4
08	Production & operationsEval gatingObservabilityDrift managementUnit economics and handover	30	45 min	4

Single best answerOne defensible option among plausible distractors drawn from real field mistakes.

Scenario judgementA client situation with constraints; you pick the action a Senior SCDAI Engineer would take.

Architecture selectionGiven non-functional requirements, choose the pattern and justify the trade-off.

Failure diagnosisA trace, an eval report or a cost curve — identify the root cause.

Each module ends with a timed assessment; the program certificate requires a pass on every one.
Items are drawn at random from a versioned bank, so no two learners sit an identical paper.
Pass mark is 70%. Two retakes are included, each with a fresh draw.
Scenario items carry double weight — they are the ones that predict field performance.
Capstones are reviewed against a published rubric by a Senior SCDAI Engineer, not auto-graded.

Try four questions from the bank

Real items, drawn from 32 in this program's bank — weighted toward the scenario and diagnosis types, because those are the ones that predict field performance. Instant feedback, nothing saved.

Four sample items — one attempt each, then the reasoning is shown.

Q1A bank asks you to "add AI to onboarding". What do you establish first?

The engineer sizes volume × error cost before choosing anything technical. That number decides whether the work is worth doing and which decision inside onboarding to target.

Q2Core banking systems are typically batch-oriented. What does that mean for your design?

Promising a real-time workflow on top of a batch core is the classic banking integration failure. Negotiate the latency budget before you promise the workflow.

Q3Your extraction agent returns confident answers for fields that are absent from the document. What is the fix?

Structural constraints beat instructions. Requiring a resolvable span makes absence detectable instead of inviting invention.

Q4A question asks which customers are exposed to a delayed counterparty. Why does vector search struggle?

Vector search finds similar text; a graph encodes who is connected to whom. Multi-hop exposure questions need traversal.

Career & outcomes

What this makes you, and where it goes next

The same ladder whichever specialisation you enter through — what changes is the domain you go deep in. Below: how the program is delivered, the skills it moves, the roles it leads to, and the specialisations closest to this one.

Available two ways

The same labs, assessments and capstones, delivered to an enterprise cohort or to individual professionals.

B2B

Direct to enterprise

Cohorts of 20 to 2,000+ on your own tenancy, with your data patterns and your cloud. Skill-gap baselining up front, per-team mastery reporting throughout, and capstones scoped against your real backlog so the output is deployable work.

Private cohort on your tenancy
Baseline skills assessment before kick-off
Capstones scoped to your backlog
Per-team mastery dashboards for L&D
Optional Senior SCDAI Engineer architecture reviews

B2C

Direct to individual professionals

The same labs, assessments and capstones for individual engineers and analysts, run on shared infrastructure with a fixed cohort calendar. You leave with a graded portfolio, not a certificate of attendance.

Fixed-date cohorts, self-paced within them
Same graded labs and item bank
Portfolio-ready capstones you own
Certificate on a pass in every module
Community and office-hours access

Skill matrix

Find your row and aim one column right. The Playground scores you against this after every module.

Skill	Beginner	Intermediate	Advanced
Banking domain fluency	Knows the products.	Maps a workflow to decisions and owners.	Sizes value and designs the operating change.
Regulated deployment	Aware of model risk.	Writes model documentation as they build.	Gets deployments through model risk first time.
Financial data	Reads a core extract.	Resolves entities and evidences lineage.	Designs the governed data path for agents.
Context engineering	Writes clear prompts.	Structures retrieval, tools and state deliberately.	Designs context strategy for reliability and cost at scale.
Retrieval & grounding	Builds basic vector search.	Tunes chunking, hybrid search and reranking.	Designs graph + vector grounding with measured recall.
Agent orchestration	Runs a single tool-calling agent.	Builds supervised multi-step and multi-agent flows.	Designs autonomy boundaries and failure containment.
Tool & system integration	Calls a documented API.	Writes an MCP server over a system of record.	Designs a least-privilege tool estate across systems.
Evaluation	Eyeballs outputs.	Builds labelled eval sets and regression gates.	Runs online evals with drift and judge calibration.
Observability & cost	Reads logs.	Traces runs, tracks tokens and latency.	Owns cost per task and capacity planning in production.
Security & guardrails	Adds output filters.	Mitigates the OWASP LLM Top 10 in a build.	Threat-models an agent estate and proves controls.
Client delivery	Takes notes in a workshop.	Runs discovery and scopes a thin slice.	Owns the account technically, from scope to handover.

Where it takes you

The SCDAI ladder is the same whichever specialisation you enter through — what changes is the domain you go deep in.

Data and AI EngineerEntry

Python, LLM APIs, first RAG build, prompt and context basics.

Associate Data and AI EngineerPractitioner

Client-facing discovery, tool integration, agent build under supervision.

SCIKIQ Certified Data and AI EngineerSpecialist

Owns a deployment end to end — scoping, build, evals, handover.

Senior Data and AI Engineer / Domain LeadSenior

Multi-agent architecture, regulated deployments, reference patterns for the practice.

AI Architect / Practice LeadLeadership

Portfolio of accounts, delivery standards, hiring and capability strategy.

Terms worth knowing

KYC / CDD: Know Your Customer and Customer Due Diligence — identity and risk verification obligations.

SAR: Suspicious Activity Report — the regulatory filing an investigation may produce.

Model risk: The governance regime (SR 11-7 style) covering validation, documentation and monitoring of models.

BCBS 239: Principles for risk data aggregation and reporting — effectively a lineage mandate.

ISO 20022: The structured messaging standard replacing legacy payment formats.

Adverse action: A declined or restricted decision that must be explained to the customer.

Straight-through processing: A transaction completed with no human touch.

SCDAI: SCIKIQ Certified Data and AI Engineer — the credential this program awards.

SCDAI engineer: An engineer embedded with the customer who turns their constraints into a working system.

Thin slice: The narrowest end-to-end path through a workflow that still produces business value.

Agentic RAG: Retrieval where the agent decides when, what and how often to retrieve.

MCP: Model Context Protocol — open standard connecting an agent to tools and data.

A2A: Agent2Agent — protocol for agent-to-agent delegation across vendors.

Eval set: A labelled, versioned set of cases an agent must pass before release.

LLM-as-judge: Using a model to score outputs, calibrated periodically against human ratings.

Guardrail: An enforced constraint on an input, output or action — not a prompt request.

Human-in-the-loop: A required human approval step before a consequential action executes.

Cost per task: Total token, tool and compute cost of one completed unit of work — the engineer's unit economic.

Golden path: The reference architecture a practice standardises on so deployments stay reviewable.

Handover: The point at which the client can operate, evaluate and change the system without the engineer.

Related programs

Skill a team, or join a cohort

B2B cohorts run on your tenancy with capstones scoped to your backlog. B2C cohorts run on a fixed calendar.

Talk to us

Enrolling

What you need, what you get, and what it costs

Stated plainly enough to rule yourself in or out without a sales call: the prerequisites, how the program runs, exactly what the credential is worth, and the questions everyone asks.

Before you start

Stated plainly so you can rule yourself in or out without a sales call. Nothing here is a formal qualification — it is what the first lab assumes you can already do.

Technical

You should already be able to do these

Comfortable in Python: functions, typing, virtual environments, reading a stack trace
Have called an LLM API and handled a structured response
Git, the command line, and reading someone else’s code without panic

Domain

What we assume, and what we teach

Know what a loan, a limit and a current account are — product depth is taught, basics are not
Any prior exposure to a regulated environment helps; it is not required

Time and access

What the program asks of your week

A machine that can run a browser and a terminal — labs run in the Playground, not locally
Roughly 8 hours a week for the taught weeks, plus 20 hours for the two capstones

How it runs

Cohort dates and pricing are confirmed on enquiry rather than printed here, because both move with the intake.

Format: Self-paced within a fixed cohort window, delivered in the SCIKIQ Agentic AI Playground
Taught hours: 80 hours across 8 modules
Capstones: 20 hours — two end-to-end builds, reviewed not auto-graded
Typical duration: 10–12 weeks at around 8 hours a week
Level: Intermediate → Advanced
Built for: Engineers, architects and analysts deploying AI inside banks
Next cohorts: — confirmed on enquiry
Pricing: — confirmed on enquiry

The credential

The credential is awarded per specialisation, so it names the domain or stack you were assessed in rather than claiming general competence. On this program the badge reads SCDAI — Banking & Financial Services.

How it is earned

A pass at 70% or above on every module assessment — not an average across them
Both capstones reviewed against the published rubric by a Senior SCDAI Engineer
All labs submitted and machine-graded green

Awarded as: A verifiable digital badge with a public credential record
Valid for: Two years from award
Recertification: One current-year assessment plus a refreshed capstone
Retakes: Two included per assessment, each with a fresh item draw
Named specialisation: The badge states the domain or stack, e.g. SCDAI — Healthcare
Verification: Third parties confirm a holder from the credential record, no login required

A certificate that cannot be checked is decoration. Every award resolves to a record showing the specialisation, the award date and the assessments passed.

The certificate you leave with

This is the credential the program awards, shown exactly as it is issued — with the specialisation named, the assessment record attached and a verification link anyone can check without an account.

SCIKIQ Data Academy Certificate of Professional Certification

This is to certify that

Your name

has been assessed and certified as

SCIKIQ Certified Data and AI Engineer

Banking & Financial Services

Specialisation: Banking & Financial Services
Assessments passed: 8 of 8, each at 70% or above
Labs graded: 40 of 40 machine-graded green
Capstones reviewed: 2, against the published rubric
Taught hours: 80 across 8 modules, plus 20 capstone hours

AwardedOn completion

Valid untilTwo years from award

Credential IDSCDAI-BANKING-FINA-<issued>

Senior SCDAI Engineer, Certification Board

01
The specialisation is on the certificate
Not “Data and AI Engineer” but the domain or stack you were actually assessed in. A general claim would be a weaker one.
02
The assessment record travels with it
Modules passed, labs graded and both capstones reviewed — so the credential states what was measured rather than that you attended.
03
Anyone can verify it, without a login
The credential ID resolves to a public record showing the specialisation, the award date and the assessments passed.

A certificate that cannot be checked is decoration. Every award resolves to a record showing the specialisation, the award date and the assessments passed.

Questions we get asked

The objections that come up in every conversation about this program, answered without the brochure voice.

Is this a course or a certification?

Both, and the second is the point. Eight timed assessments and two reviewed capstones stand between you and the credential, so a pass means someone measured the skill rather than recorded your attendance.

How is it self-paced without falling apart?

Every lab is provisioned, graded and unblocked by the Playground rather than by an instructor. That is what lets a cohort of 2,000 cost the same faculty time as a cohort of 20 — and why you are never waiting on someone to mark your work.

What if I fail an assessment?

Two retakes are included per assessment, each drawing a fresh item set from the bank, so a retake is a genuinely new paper rather than the same questions again.

Do I need my own cloud account?

For the domain specialisations, no — labs run in provisioned sandboxes. For the four tech-stack programs you will want access to that platform, since deploying into a real subscription is much of the point.

Can we run this privately for our team?

Yes. B2B cohorts run on your own tenancy with your data patterns, a skills baseline before kick-off, per-team mastery reporting, and capstones scoped against your actual backlog so the output is deployable work rather than an exercise.

Which specialisation should I take?

Take the domain you deploy into. If you move across industries, take a tech-stack program instead and pick up domain context on the engagement. The chooser on the programs page will narrow it.

How current is the material?

The trends, platform capabilities and regulatory positions are reviewed each quarter, and every external claim on these pages links to its source so you can check the date yourself.

What do I actually leave with?

A graded portfolio: forty machine-graded labs, two reviewed end-to-end agent builds with measured evaluation and cost per task, and a verifiable credential naming your specialisation.

Still deciding?

Tell us the systems you deploy into and we will say plainly whether this specialisation is the right one — or which of the 21 is.

Talk to us

SCIKIQ Certified Data and AI EngineerBanking & Financial Services

The value chain, the people, the trends

The banking value chain

Acquire & onboard

Decisions made here

Where it breaks today

What is measured

Underwrite & originate

Decisions made here

Where it breaks today

What is measured

Service & operate

Decisions made here

Where it breaks today

What is measured

Detect & protect

Decisions made here

Where it breaks today

What is measured

Manage risk & capital

Decisions made here

Where it breaks today

What is measured

Report & comply

Decisions made here

Where it breaks today

What is measured

Who you are actually building for

What the strategy houses are seeing

Operations is the value pool, not the demo

Frontline rewiring pays in months

The headline number is productivity

Redesign the process, do not decorate it

Concentrate, do not sprinkle

Trust is the deployment constraint

Domains, ontology, taxonomy and governed terms

The data landscape you will actually meet

Core banking

Payments & messaging

Customer & CRM

Risk & finance marts

Unstructured corpus

Surveillance & case

The semantic layer

Data domains

Party

Arrangement

Transaction

Risk & exposure

Financial crime

Product & policy

Ontology & knowledge graph

Party

Customer

Account

Facility

Collateral

Payment

Alert

Case

Exposure

Rating

Product

Policy

Taxonomies

Product hierarchy

Customer risk rating

Financial crime typology

Governed business terms

How it is actually built

The reference architecture

Sources & systems of record

Ingestion & CDC

Governance & semantics

Retrieval & grounding

Agent & orchestration

Guardrails & model risk

Evaluation & observability

Typical tech stack

Models