IndustryAssurance first
Assurance is where agentic value lands first
Alarm correlation and customer-impact scoping are bounded, high-volume and measurable — the profile that agentic automation handles well and executives can verify.
Program 13 · Domain specialisation
SCIKIQ Certified Data and AI EngineerTelecom
A telco knows everything about its network and almost nothing about what that network did to a particular customer on Tuesday. The two halves of the estate — OSS and BSS — were built by different people for different reasons and join badly. This specialisation is about closing that join, because nearly every worthwhile use case sits across it.
- 80 hrstaught
- 40hands-on AI labs
- 8assessment tests
- +20 hrs2 E2E capstones
OSS × BSSThe join where almost all operator value sits
Care & assuranceWhere agentic automation lands first and most reliably
Revenue assuranceA leakage problem that is a data-reconciliation problem
Capex efficiencyNetwork investment steered by experience, not coverage maps
01
Part 1 · The business
The value chain, the people, the trends
An operator is two businesses sharing a brand: one runs a national engineering asset, the other runs a mass-market subscription. Part 1 maps both and names the people who work across the seam.
The operator value chain
Six stages from spectrum to retention. Select one.
01
Plan & build
Spectrum, radio planning, rollout, densification and capex allocation.
Decisions made here
- Where do we build next?
- Is this site justified?
- What do we upgrade first?
Where it breaks today
- Coverage maps that do not reflect experienced quality
- Rollout decisions made on planning tools, not on customer impact
- Capex cases that cannot be traced to a churn or revenue outcome
What is measured
- Capex per site
- Population coverage
- Capacity headroom
Where AI landsSite prioritisation from experienced quality, permit-pack drafting and rollout risk flagging.
Capex per sitePopulation coverageCapacity headroom
02
Operate & assure
Network operations, fault management, performance and service assurance.
Decisions made here
- What is actually broken?
- Who is affected?
- What do we fix first?
Where it breaks today
- Alarm storms in which one root fault raises four hundred alarms
- Correlation done by the two engineers who remember the network
- Degradation that never alarms but customers feel
What is measured
- MTTR
- Availability
- Customer-affecting incidents
Where AI landsAlarm correlation to a single root cause, customer-impact scoping and dispatch triage.
MTTRAvailabilityCustomer-affecting incidents
03
Acquire & provision
Sales, order capture, credit, activation and service provisioning.
Decisions made here
- Can we serve this address?
- Is this order clean?
- Why did activation fail?
Where it breaks today
- Serviceability answers that are wrong at the address level
- Order fallout worked manually by a back-office team
- Provisioning failures diagnosed by trial and error
What is measured
- Order fallout rate
- Time to activate
- First-time-right provisioning
Where AI landsFallout diagnosis and auto-correction, serviceability grounding and activation triage.
Order fallout rateTime to activateFirst-time-right provisioning
04
Serve & support
Contact centre, digital care, field service and complaint handling.
Decisions made here
- What is this customer experiencing?
- Can we fix it remotely?
- Do we need to roll a truck?
Where it breaks today
- Agents who cannot see the network state behind the complaint
- Truck rolls dispatched for faults that were never in the home
- Repeat contacts because the first answer addressed the symptom
What is measured
- Contacts per subscriber
- Truck-roll rate
- First contact resolution
Where AI landsNetwork-aware care answers, truck-roll avoidance and complaint-to-fault linking.
Contacts per subscriberTruck-roll rateFirst contact resolution
05
Rate, bill & assure revenue
Mediation, rating, billing, collections and revenue assurance.
Decisions made here
- Was this usage billed correctly?
- Where are we leaking?
- Is this dispute justified?
Where it breaks today
- Mediation and rating quietly disagreeing on new plans
- Leakage found in an annual audit rather than continuously
- Bill-shock disputes handled one at a time
What is measured
- Revenue leakage
- Dispute volume
- Days sales outstanding
Where AI landsContinuous usage-to-bill reconciliation, leakage detection and dispute explanation.
Revenue leakageDispute volumeDays sales outstanding
06
Retain & grow
Churn management, upsell, loyalty and lifecycle marketing.
Decisions made here
- Who is about to leave?
- What offer holds them?
- Is this worth saving?
Where it breaks today
- Churn models that fire after the customer has decided
- Save offers made without knowing the customer had a network problem
- Consent state that is checked at send time, too late to matter
What is measured
- Churn rate
- ARPU
- Save rate
Where AI landsExperience-grounded churn signals, next-best-action and consent-safe outreach.
Churn rateARPUSave rate
Who you are building for
In telecom the NOC engineer and the care agent are looking at the same fault and cannot see each other. Select one to light the stages they own.
What the sector is seeing
An industry with excellent data, poorly joined, and a cost base that rewards joining it.
IndustryNetwork × customer
The OSS/BSS join is the differentiator
Operators that can say which customers a given fault affected outperform those that can describe the fault perfectly and the impact not at all.
IndustryAvoidable dispatch
Truck rolls are the visible cost line
Dispatch is the most expensive routine action an operator takes, and a large share of visits address faults that were never inside the home.
IndustryContinuous, not annual
Revenue leakage is a reconciliation problem
Leakage is rarely fraud and usually disagreement between mediation, rating and billing on new plans — a continuous data problem dressed as a finance one.
IndustryConsent at retrieval
Consent moved upstream
Marketing permission and lawful basis now have to be enforced where data is retrieved, not where a message is sent, which changes the architecture rather than the process.
IndustryExperience-led capex
Network capex is being steered by experience
Investment decisions are shifting from coverage modelling toward measured customer experience, which requires the two data estates to be joined first.
02
Part 2 · Data and meaning
Domains, ontology, taxonomy and governed terms
The operator estate is split down the middle. OSS knows the network, BSS knows the customer, and the identifiers that should join them — cell, circuit, service, subscriber — do not agree. Most of the engineering here is closing that gap credibly.
The data landscape
Six families. The join keys between the first three and the last three are the whole problem.
Network fault & performance
Alarms, KPIs and counters from RAN, core and transport.
EMS/NMS, performance managersNetwork inventory & topology
Sites, cells, circuits, ports and the paths between them.
Inventory systems, GISUsage records
Call, data and messaging records through mediation, at very high volume.
Mediation platforms, CDR/EDR storesCustomer & order
Subscribers, products, orders, provisioning state and entitlements.
CRM, order managementBilling & revenue
Rating, invoices, adjustments, disputes and collections.
Billing platforms, revenue assuranceInteraction & field
Contact transcripts, tickets, work orders and field notes.
Contact centre, ticketing, field serviceThe semantic layer
An operator agent that cannot tell a cell from a service, or a subscriber from an account, will describe a fault perfectly and name the wrong customers. These are the domains, the graph and the definitions the agent must be grounded on before it is allowed near a care conversation.
Data domains
Six subject areas, each with the entities it holds, its critical data elements and the function accountable for it.
Network asset
The physical and logical estate that carries traffic.
SiteCellNodeCircuitPort
- Critical data elements
- Cell identifier; Circuit identifier; Functional location
- Owned by
- Network planning
Service
What the customer bought, expressed in terms the network can resolve.
ServiceService instanceProductEntitlement
- Critical data elements
- Service identifier; Service class; Contracted rate
- Owned by
- Product & service management
Customer & account
Who is billed, who uses the service, and the difference between them.
CustomerAccountSubscriberContact
- Critical data elements
- Account number; Subscriber identifier; Consent state
- Owned by
- Customer data
Usage
Consumption records produced by the network and mediated for rating.
Usage recordSessionRated event
- Critical data elements
- Usage volume; Event timestamp; Rating key
- Owned by
- Mediation & revenue assurance
Fault & incident
What broke, when, and which services it took with it.
AlarmIncidentOutageTicket
- Critical data elements
- Severity; Start and clear time; Affected service count
- Owned by
- Network operations
Billing & revenue
Rating, invoicing, adjustments and the disputes they generate.
InvoiceChargeAdjustmentDispute
- Critical data elements
- Invoice amount; Tariff applied; Adjustment reason
- Owned by
- Billing & finance
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
Core entity
Customer
The legal entity or person who holds the account.
- Customer holds Account
Core entity
Account
The billing relationship.
- Account billed via Invoice
- Account contains Subscriber
- Customer holds Account
Core entity
Subscriber
The identity that consumes the service.
- Subscriber consumes Service
- Subscriber generates Usage record
- Account contains Subscriber
Core entity
Service
The instance of a product delivered to a subscriber.
- Service instantiates Product
- Service carried by Circuit
- Service served by Cell
- Subscriber consumes Service
- Incident affects Service
Reference / classification
Product
The catalogue offering a service instantiates.
- Service instantiates Product
Core entity
Circuit
The logical path carrying the service.
- Service carried by Circuit
Core entity
Cell
The radio cell serving an area.
- Cell hosted at Site
- Service served by Cell
- Alarm raised by Cell
Core entity
Site
The physical location the cell sits on.
- Cell hosted at Site
Event
Alarm
A fault signal raised by network equipment.
- Alarm raised by Cell
- Alarm correlates to Incident
Event
Incident
The correlated root fault behind many alarms.
- Incident affects Service
- Alarm correlates to Incident
Event
Usage record
A mediated record of consumption.
- Usage record rated onto Invoice
- Subscriber generates Usage record
Event
Invoice
The billed statement for an account.
- Account billed via Invoice
- Usage record rated onto Invoice
Taxonomies
The classification hierarchies that make records comparable across systems.
Network hierarchy
- Region
- Site
- Node
- Cell
West → Site 4412 → eNodeB 88 → Cell 88-2
Fault severity
- Class
- Severity
- Customer impact
Service affecting → Critical → 3,100 subscribers
Product hierarchy
- Family
- Product
- Plan
- Add-on
Mobile → Postpaid → Unlimited 5G → International roaming
Governed business terms
The definitions an agent must use rather than invent. Most wrong answers in this industry are a term used loosely.
- Customer vs subscriber
- The customer is billed; the subscriber consumes. One account routinely holds many subscribers, and conflating them misdirects both care and marketing.
- Service affecting
- A fault that degrades a service a customer bought, as distinct from one that degrades the network without reaching a customer.
- Customer affected
- The agreed count of subscribers whose service an incident degraded — needs one definition, or network and care will report different numbers for the same event.
- Alarm vs incident
- An alarm is a signal; an incident is the correlated fault behind many of them. Counting alarms as incidents overstates severity.
- Revenue leakage
- Service delivered but not correctly billed, usually from mediation and rating disagreeing rather than from fraud.
- Order fallout
- An order that stalls in provisioning and requires manual correction before service starts.
- Truck roll
- Dispatch of a field engineer — the most expensive routine action in the operator cost base.
- MTTR
- Mean time to repair, measured from incident detection to service restoration rather than to alarm clear.
03
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, with the OSS/BSS join treated as a first-class governed entity.
01/07
OSS, BSS & the join
Two estates read through their own paths, plus the inventory that maps one to the other.
Takes in
Nothing. This is the upstream edge.
Hands on
Alarm and usage streams, topology snapshots and customer state, each with its own freshness.
SystemWhat it doesKey business entities
Fault and performance management
Alarms, counters and KPIs from radio, core and transport. Raises hundreds of signals for one underlying fault, which is why correlation happens before anything else reads it.
AlarmIncidentSeverity
Network inventory and topology
Sites, cells, nodes, circuits and ports, and the paths between them. The only thing that turns a stream of alarms into a causal story, and it is never fully complete.
SiteCellCircuitPort
Mediation
Collects, normalises and forwards usage records for rating. Very high volume, and the point where usage and billing first have the chance to disagree.
Usage recordSessionRating key
CRM and order management
Subscribers, products, orders and provisioning state. Where an order stalls, and where the customer-side identity that must join to the network lives.
CustomerAccountSubscriberService
Billing and revenue assurance
Rating, invoicing, adjustments and disputes. Holds the tariff that produced a charge, which is what any dispute explanation has to be grounded on.
InvoiceChargeAdjustmentDispute
Contact centre and field service
Tickets, transcripts, work orders and engineer notes — the largest unstructured corpus an operator owns, and the record of what actually fixed a fault.
TicketConversationWork order
What the engineer owns here
Establishing which identifier actually joins a cell to a subscriber, and how stale it is.
Where this layer fails
Assuming the subscriber-to-cell mapping is a key. It is inferred, it moves constantly, and any design that treats it as exact will name the wrong customers with total confidence.
02/07
Ingestion & correlation
High-volume streaming ingestion with topology-aware correlation applied at the boundary.
Takes in
Read-mostly access paths, change feeds and document streams from the systems of record.
Hands on
Replayable, resolved, quality-checked records with their lineage back to the source row.
SystemWhat it doesKey business entities
Change data capture
Reads the source’s own change log rather than polling it, so the platform sees every state a record passed through instead of only where it ended up.
Change eventOperation typeCommit timestamp
Landing and raw zone
Every extract kept as it arrived, immutable and timestamped. The thing you replay from when a downstream definition turns out to have been wrong.
Raw extractBatchWatermark
Document processing
Layout-aware extraction over the unstructured half of the estate: sectioning, tables, signatures, and the page reference every later citation depends on.
DocumentSectionPage reference
Entity resolution
Decides that two records are the same real-world thing, with a survivorship rule and a confidence, so downstream joins are a decision rather than an assumption.
Match candidateSurvivorship ruleCluster
Data quality and contracts
Schema, freshness and volume expectations asserted at the boundary, so a bad load fails loudly here rather than quietly two layers later.
ContractExpectationViolation
Topology enrichmentthis domain adds
Attaches every alarm to the inventory path it sits on, which is what turns a stream of signals into a causal story rather than a list.
AlarmPathCircuit
What the engineer owns here
Reducing an alarm storm to candidate root causes before anything downstream sees it.
Where this layer fails
Correlating on time order. The first alarm to arrive is rarely the cause, and a system that assumes it is will confidently blame the symptom.
03/07
Service model & semantics
The service graph: subscriber to service to circuit to cell to site, with governed metrics.
Takes in
Replayable, resolved records with lineage from ingestion.
Hands on
Typed, classified, defined data an agent can be grounded on without inferring meaning.
SystemWhat it doesKey business entities
Business glossary
One agreed definition per term, owned by a named person, so the number an agent quotes means what the business means by it.
TermDefinitionOwnerSteward
Ontology and knowledge graph
The typed entities and relationships the domain actually has, so an agent can traverse "which customers are exposed to this" rather than guess from adjacent text.
EntityRelationshipNodeEdge
Semantic and metrics layer
Metrics defined once, in one place, with their filters and grain. Removes the class of error where the agent computed something plausible and wrong.
MetricDimensionGrainFilter
Classification and policy
Sensitivity labels and the purpose each classification permits, applied at the field level and inherited by everything downstream.
ClassificationPurposePolicy
Lineage and catalogue
Where a value came from and what it touched on the way. The layer that makes an answer defensible rather than merely correct.
Lineage edgeDatasetColumn
What the engineer owns here
One definition of "customer affected", agreed by network and by care.
Where this layer fails
Letting the model infer meaning from column names. It will, it will be plausible, and nobody will notice until the number reaches a regulator or a board pack.
04/07
Retrieval & grounding
Retrieval over runbooks, vendor documentation, tickets and prior incident resolutions.
Takes in
Typed, classified, defined data from the governance layer.
Hands on
Ranked, permission-trimmed, citable evidence scoped to the caller and the moment.
SystemWhat it doesKey business entities
Hybrid index
Vector and lexical retrieval together, because exact identifiers, codes and clause numbers are the thing semantic search is worst at.
ChunkEmbeddingTermScore
Chunking and metadata
Splits on the document’s own structure and carries effective dates, version and source into every chunk, so a retrieved passage knows when it was true.
ChunkEffective dateVersionSource reference
Security trimming
Applies the caller’s permissions inside the query rather than filtering results afterwards, so the model never sees what the user may not.
PrincipalEntitlementFilter
Graph retrieval
Traverses the knowledge graph for questions that are joins rather than similarity — exposure, genealogy, ownership, causation.
PathHopTraversal
Reranking and citation
Reorders candidates on relevance and returns the passage identifier behind every sentence, so the answer can be checked rather than trusted.
CandidateRankCitation
What the engineer owns here
Grounding a diagnosis in the runbook and the prior ticket that actually fixed it.
Where this layer fails
Filtering results after retrieval instead of constraining the query. The model has already seen the rows you removed, and it will use them.
05/07
Agent & orchestration
Agents for alarm triage, care diagnosis, dispatch avoidance and billing reconciliation.
Takes in
Ranked, permission-trimmed, citable evidence from retrieval.
Hands on
Actions taken or proposed, each with the evidence, the identity and the trace behind it.
SystemWhat it doesKey business entities
Agent runtime
Plans, calls tools and holds the loop. Where the step budget, the timeout and the stopping condition are enforced rather than hoped for.
TaskPlanStepTrace
Tool and MCP servers
Typed, permissioned tools with declared schemas. An agent’s real capability surface is this list, which is why the list is a security artefact.
ToolSchemaScopeInvocation
State and memory
Durable task state with checkpoints, so a run that dies mid-way resumes instead of restarting and re-doing side effects.
CheckpointSessionMemory record
Human-in-the-loop gates
Approval steps on the actions that need one, carrying enough context for the approver to actually decide rather than rubber-stamp.
ApprovalThresholdDecision
Workflow integration
Writes back into the systems people already work in, under a service identity with its own audit trail.
Work itemWrite-backActor
What the engineer owns here
The autonomy line: which actions the agent takes and which it only recommends.
Where this layer fails
An agent that acts under a service account rather than on behalf of the user. It will eventually do something the requesting user had no right to do, and the log will not show it.
06/07
Consent, guardrails & regulation
Consent enforcement, lawful intercept boundaries, retention limits and audit.
Takes in
Proposed actions and drafted responses from the agent layer.
Hands on
Permitted, grounded, logged output — or a refusal with a stated reason.
SystemWhat it doesKey business entities
Policy enforcement
The rules that decide whether an action is permitted at all, evaluated before the action and independently of the model that proposed it.
PolicyDecisionObligation
Input and output screening
Injection detection on the way in, and on the way out the checks for leakage, unsupported claims and content the domain forbids.
PromptResponseViolation
Groundedness checking
Verifies each assertion resolves to retrieved evidence, and fails the response rather than shipping the sentence that does not.
AssertionSupportVerdict
Model risk management
Model inventory, intended use, validation evidence and the sign-off that lets a model be used for a purpose. Not optional in a regulated estate.
ModelIntended useValidationApproval
Audit log
Immutable record of what was asked, retrieved, decided and done — the artefact a reviewer reads when they do not take your word for it.
Audit entryActorEvidence reference
Purpose and retention controlthis domain adds
Enforces the purpose communications metadata was collected under, and the retention limit, inside the index rather than only in the source system.
PurposeRetention windowLawful basis
What the engineer owns here
Ensuring usage data is used for the purpose it was collected under, and no other.
Where this layer fails
Reusing usage records for a purpose they were not collected under. It is a lawful-basis question, and it is settled before the architecture, not after the pilot.
07/07
Evaluation & observability
Correlation accuracy, false-dispatch rate, drift and cost per subscriber.
Takes in
Permitted, grounded, logged output from the guardrail layer.
Hands on
Measured quality, cost and latency — and the evidence to change any of the three.
SystemWhat it doesKey business entities
Evaluation harness
Held-out sets and graded runs on every change, so a prompt edit is a measured change rather than a hopeful one.
Eval setCaseScoreBaseline
Regression gate
Blocks a release when quality drops, in CI, on the same evidence for everyone. The difference between a system and a demo.
ThresholdRunGate decision
Tracing
End-to-end spans across retrieval, model and tool calls, so a bad answer can be opened and read rather than argued about.
TraceSpanLatencyToken count
Cost and capacity
Cost per task and per tenant, against throughput and quota. The number that decides whether the pilot can become the rollout.
Cost per taskQuotaThroughput
Drift and feedback
Watches quality against production traffic rather than the test set, and routes real corrections back into the eval suite.
Drift signalFeedbackIncident
What the engineer owns here
Proving the agent reduced truck rolls without increasing repeat faults.
Where this layer fails
Evaluating once, before launch. Quality moves with the data, the model and the traffic, and a system with no live measurement has no idea which of the three moved.
Typical tech stack
Models
- Frontier models for diagnosis and explanation
- Small models for high-volume triage
- Anomaly models on counters
Grounding
- Runbook and vendor document index
- Ticket resolution history
- Service knowledge graph
Agents
- Alarm triage agents
- MCP over inventory and ticketing
- Care assist with network context
Data
- Streaming ingestion at CDR volume
- Topology enrichment
- OSS/BSS identity resolution
Assurance
- Correlation accuracy evals
- False-dispatch measurement
- Consent and retention controls
Runtime
- Multi-region deployment
- Read paths into OSS
- Cost per subscriber budgeting
Constraints that shape the design
Volume is genuinely large
Usage records arrive at a scale where per-record model inference is economically impossible. Triage cheaply, reason expensively, and know which is which.
The join is imperfect and always will be
Subscriber-to-cell mapping is probabilistic and stale. Design for confidence-scored joins rather than pretending to a clean key.
Usage data is regulated
Communications metadata carries purpose limitation and retention rules. Reuse for a new purpose is a legal question before it is a technical one.
A wrong dispatch costs real money
False positives are not a quality metric here; each one is a van, an engineer and a customer appointment.
04
Part 3 · Getting it done
Where the business and the technology meet
Every use case below crosses the OSS/BSS seam, because that is where an operator has advantages nobody outside the industry can copy.
Use cases, mapped to the chain and the architecture
Filter by stage or by earned autonomy. 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
No use cases match that combination.
The delivery playbook
How a telecom deployment actually runs.
01
Week 1 — test the join before promising anything
Measure how well subscriber, service, circuit and cell actually reconcile. Every use case downstream is capped by this number, so discover it before it is a commitment.
02
Week 2 — agree what "customer affected" means
Get network and care to sign one definition. Without it the same incident will carry two severities and the agent will be blamed for both.
03
Weeks 3-6 — start in assurance, not in care
Alarm correlation is bounded, measurable and touches no customer. It builds the topology confidence that every customer-facing use case then depends on.
04
Weeks 7-9 — take on dispatch with the asymmetry modelled
Truck-roll avoidance is the value case, but a wrong "no dispatch" costs more than a wasted visit. Set thresholds with field operations, not with data science.
05
Weeks 10-12 — consent, retention and handover
Prove usage data was used only for its collected purpose, then hand over runbook, thresholds and the reconciliation evidence.
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 an operatorLanding in an organisation where network and commercial report to different worlds.10 hrs · 5 labs
Delivery mandateOSS/BSS organisational splitValue sizing on dispatch and churnStakeholder mapAdoption dynamicsThin-slice scoping
01
Decision selectionScore four operator decisions on value, cadence and data readiness.
02
Value sizingSize a truck roll, a repeat contact and a churn event in cash terms.
03
Join auditAssess whether subscriber-to-cell data supports a customer-impact use case.
04
Stakeholder mapMap NOC, care, field, revenue assurance and network planning.
05
Thin sliceScope a single-region assurance slice with a measurable outcome.
Assessment 01 — Operating model & scoping
02Operator data foundationsFault, performance, inventory, mediation, CRM and billing — and the keys between them.10 hrs · 5 labs
Alarm and counter dataInventory and topologyMediation and usage recordsCRM and order stateBilling dataIdentity resolution across OSS and BSS
01
Alarm ingestionIngest and de-duplicate a realistic alarm stream.
02
Topology enrichmentEnrich alarms with the inventory path they sit on.
03
Usage at volumeAggregate high-volume usage records within a cost budget.
04
Join measurementMeasure subscriber-to-cell join quality and report it honestly.
05
Billing extractReconcile a rated usage sample against the issued invoice.
Assessment 02 — Operator data landscape
03Context engineering for network languageVendor shorthand, cell identifiers, severity scales and the precision they demand.10 hrs · 5 labs
Identifier disambiguationVendor terminologyStructured outputsConfidence and uncertaintyRefusal on insufficient topologyContext budget
01
Identifier resolutionResolve ambiguous cell and circuit references to inventory.
02
Vendor normalisationNormalise alarm text across two vendors to one vocabulary.
03
Structured diagnosisEmit a structured root-cause hypothesis with evidence and confidence.
04
Honest degradationMake the agent state reduced confidence when topology is incomplete.
05
Context budgetCut context cost on a high-frequency triage agent.
Assessment 03 — Context engineering
04Grounding on runbooks and ticket historyThe fix for this fault has almost certainly been written down already.10 hrs · 5 labs
Runbook retrievalVendor documentationTicket resolution miningService graph traversalCitationsRetrieval evals
01
Runbook indexBuild retrieval over runbooks with procedure and revision cited.
02
Ticket miningExtract what actually resolved a fault from free-text ticket history.
03
Service traversalTraverse subscriber to service to circuit to cell in the graph.
04
Citation contractCite the runbook step and the prior ticket behind a recommendation.
05
Retrieval evalsBuild a labelled fault-question set and measure recall.
Assessment 04 — Retrieval & grounding
05Assurance and care agent designCorrelation, impact scoping and the dispatch decision, with the cost asymmetry modelled.10 hrs · 5 labs
Workflow vs agentAlarm correlation patternsImpact scopingDispatch decisioningAsymmetric costEscalation
01
Correlation agentCollapse an alarm storm to a ranked root-cause set.
02
Impact scopingAnswer who is affected in service terms with a confidence bound.
03
Dispatch decisionDecide network, line or home — and model the cost asymmetry.
04
Care assistGive a care agent network context inside a live-conversation latency budget.
05
EscalationDefine when the agent must hand to a human and prove it does.
Assessment 05 — Agent design for assurance & care
06Integrating with OSS and BSSRead paths into the network estate and write paths into ticketing and provisioning.10 hrs · 5 labs
MCP over inventory and ticketingStreaming integrationProvisioning write-backRate limits and back-pressureIdentityError handling
01
MCP over inventoryExpose topology and service lookups as typed read tools.
02
Ticket write-backCreate and update tickets under a traceable service identity.
03
Provisioning correctionApply an order-fallout fix behind an approval gate.
04
Back-pressureHandle a source system that rate-limits mid-task without losing work.
05
IdentityBind agent access to region and role entitlements.
Assessment 06 — Systems integration
07Consent, retention and telecom regulationCommunications metadata is regulated data, and reuse is a legal question first.10 hrs · 5 labs
Purpose limitationRetention limitsConsent enforcement at retrievalLawful intercept boundariesOWASP LLM Top 10Incident response
01
Purpose enforcementConstrain usage-record retrieval to its collected purpose.
02
RetentionEnforce retention limits inside the index, not only in the source.
03
Consent at retrievalApply suppression and lawful basis inside the query.
04
Injection defenceDefend against instructions embedded in a customer complaint.
05
Red teamRun an OWASP LLM Top 10 pass over the care assist pipeline.
Assessment 07 — Consent, retention & security
08Production, false dispatch and cost per subscriberAt operator volume, cost per task is not a footnote — it decides feasibility.10 hrs · 5 labs
Eval suitesFalse-dispatch measurementDriftThreshold tuningCost per subscriberHandover
01
Eval gateGate deployment on correlation accuracy regression.
02
False-dispatch studyMeasure and reduce false "no dispatch" decisions with field ops.
03
Drift watchDetect drift after a network upgrade changes alarm patterns.
04
Cost per subscriberModel and reduce inference cost across the subscriber base.
05
HandoverProduce runbook, threshold guide and reconciliation evidence.
Assessment 08 — Production & economics
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 — Assurance triage agent
Build an agent that collapses an alarm storm to a ranked root cause, scopes customer impact in service terms with a stated confidence, and cites the runbook and prior ticket behind its recommendation. Measure correlation accuracy against a labelled incident set.
Capstone B — Truck-roll avoidance agent
Build an agent that decides network, line or home before dispatch, with the cost asymmetry between a wasted visit and a missed fault modelled explicitly, and a measured false "no dispatch" rate.
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 05 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 05 · Assurance and care agent design
Alarm storm to single root cause
Collapse four hundred alarms into the one fault that caused them, and name who it hurt.
The brief
You are given a replayed alarm storm from a regional outage, the inventory topology for that region, and a labelled answer key naming the true root fault and the affected services. Topology coverage is deliberately incomplete: 18% of the paths are missing. Return the ranked root-cause candidates, the services affected by each, and a confidence that reflects the missing topology — rather than a confident answer that quietly ignores it.
Provisioned for you
- Replayed storm: 412 alarms over 90 minutes
- Inventory topology with 18% of paths absent
- Service-to-circuit-to-cell mapping
- Labelled key: 1 root fault, 3,100 affected subscribers
You start from this
def correlate(alarms, topology, services):
"""Return [(root_alarm_id, affected_service_ids, confidence)] ranked.
Topology is incomplete. Where a path is missing, say so in the
confidence; do not infer the edge and present the result as certain.
"""
# 1. group alarms by the topology path they sit on
# 2. the root is the upstream-most alarm on a path, not the first in time
# 3. walk service -> circuit -> cell to name who is affected
raise NotImplementedErrorWhat the grader asserts
- Identifies the true root fault as the top-ranked candidate
- Does not report the earliest alarm as the root — time order is not causal order
- Affected-subscriber count is within 5% of the labelled 3,100
- Confidence is materially lower for candidates whose path crosses missing topology
- Returns an explicit unknown rather than a guess when no path can be resolved at all
05
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 & scopingDelivery mandateOSS/BSS splitValue sizingThin slicing | 25 | 35 min | 3 |
| 02 | Operator data landscapeFault and performanceInventory and topologyUsage recordsOSS/BSS joins | 25 | 35 min | 3 |
| 03 | Context engineeringIdentifier resolutionVendor normalisationStructured outputsUncertainty | 25 | 35 min | 2 |
| 04 | Retrieval & groundingRunbook retrievalTicket miningGraph traversalRetrieval metrics | 25 | 40 min | 2 |
| 05 | Agent design for assurance & careCorrelationImpact scopingDispatch decisioningEscalation | 25 | 40 min | 2 |
| 06 | Systems integrationMCP designStreaming integrationWrite-back safetyIdentity | 25 | 35 min | 2 |
| 07 | Consent, retention & securityPurpose limitationRetentionConsent enforcementLLM security | 30 | 45 min | 2 |
| 08 | Production & economicsEval gatingFalse dispatchDrift and thresholdsCost per subscriber | 30 | 45 min | 2 |
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 18 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.
Q1An operator asks you to "use AI to cut truck rolls". What do you establish first?
The value case is volume × avoidable share × cost per visit. Until you have that number you cannot tell whether this is worth doing or which part of the dispatch decision to target.
Q2Usage records arrive at a volume where per-record model inference is uneconomic. What is the right pattern?
Cheap triage, expensive reasoning. Sampling loses the rare events that matter, and batching everything still pays per record for records that needed no reasoning at all.
Q3Two vendors report the same fault with different alarm text. Where should normalisation happen?
Normalising at ingestion makes the vocabulary testable and reusable. Doing it in the prompt makes every downstream agent carry the same translation burden and drift apart.
Q4A truck-roll avoidance agent must weigh a wasted visit against a missed fault. How should that be handled?
The two errors have very different costs and the ratio is a business decision, not a modelling one. Optimising raw accuracy silently picks a threshold nobody agreed to.
06
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 |
|---|---|---|---|
| Operator domain fluency | Knows the segments. | Maps network events to customer outcomes. | Sizes dispatch and churn value credibly. |
| OSS/BSS data | Queries one estate. | Joins subscriber to cell with measured confidence. | Designs an identity-resolution layer that degrades honestly. |
| Assurance engineering | Reads alarms. | Correlates a storm to a root cause using topology. | Runs correlation as a measured, gated production system. |
| 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.
01
Data and AI EngineerEntry
Python, LLM APIs, first RAG build, prompt and context basics.
02
Associate Data and AI EngineerPractitioner
Client-facing discovery, tool integration, agent build under supervision.
03
SCIKIQ Certified Data and AI EngineerSpecialist
Owns a deployment end to end — scoping, build, evals, handover.
04
Senior Data and AI Engineer / Domain LeadSenior
Multi-agent architecture, regulated deployments, reference patterns for the practice.
05
AI Architect / Practice LeadLeadership
Portfolio of accounts, delivery standards, hiring and capability strategy.
Terms worth knowing
- OSS / BSS
- Operations Support Systems (the network) versus Business Support Systems (the customer).
- CDR / EDR
- Call and Event Detail Records — the usage data mediation produces.
- Mediation
- The layer that collects, normalises and forwards usage records for rating.
- MTTR
- Mean Time To Repair — the headline assurance metric.
- Truck roll
- Dispatching a field engineer; the most expensive routine operator action.
- Revenue leakage
- Service delivered but not correctly billed.
- RAN
- Radio Access Network — the cell sites and radios customers connect to.
- Order fallout
- An order that stalls in provisioning and needs manual correction.
- 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.
07
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 roughly what a cell, a circuit and a subscriber are — the rest is taught
- Comfort with high-volume data; usage records arrive at a scale that changes design choices
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 and analysts deploying AI inside network and customer operations
- 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 — Telecom.
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
Telecom
- Specialisation
- Telecom
- 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-TELECOM-<issued>
Senior SCDAI Engineer, Certification Board
- 01The 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.
- 02The 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.
- 03Anyone can verify it, without a login
The credential ID resolves to a public record showing the specialisation, the award date and the assessments passed.
Share it on LinkedIn
Add it to your LinkedIn profile in one step. The link pre-fills the certification fields from the credential record, so the entry on your profile matches the record a reader can check.
Add to LinkedIn profile The button is live on your real certificate; here it opens LinkedIn pre-filled with this specialisation so you can see exactly what the profile entry will say.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.