Data Intelligence · Analytics

Analytics Engineer

Own the semantic model so the same question gives the same answer on different days. Show what breaks before a definition changes, not after. Keep every definition owned, tested and traceable to its source without approving its own changes.

Add this role to my team See its 8 analyses

8named analyses

21governed capabilities

8operational responsibilities

1shared Core resolver

Questions this role answers

Choose the role by the work—not by a title alone.

What breaks if this definition changes?
Where does this company disagree with itself?
Is the data still what we think it is?
Where did this column come from?
Did the source change under us?
Is the data arriving when the reports assume it does?
What is this estate costing to run?
Which models have no test behind them?

What this role actually does

One responsibility for every named analysis.

The Role Pack contains the method. The canonical profile contains the governed capability boundary. They reference each other without duplicating ServAI Core.

01
Trace every asset citing a definition and state the count before any change is proposed.
02
Find terms resolving to more than one definition and present the conflict to their owners.
03
Maintain data-quality expectations per model and report attainment against them.
04
Provide column-level lineage and name where a trace stops rather than presenting it as complete.
05
Watch source schemas for drift and report additions, removals and type changes for a person to accept.
06
Report freshness attainment against each source's declared commitment.
07
Report query cost by model, asset and owner with scheduled runs separated from human ones.
08
Report test coverage weighted by how many published assets depend on each model.

Named analytical methods

8 analyses with evidence gates and refusal conditions.

Every method maps to a registered capability the profile can actually prepare or recommend. Repeating a precondition sentence is not evidence.

01engineering.semantic.model_change

Semantic · Model Change

Trace every dashboard, report, alert and saved analysis citing the definition, count them, and present the count before any change.

Evidence gatea citation index exists

Refusesto change a definition without stating the count affected

semantic.proposal4 declared outputsconfidential

02engineering.definition.consolidation

Definition · Consolidation

Find terms resolving to more than one definition and present them side by side with their owners and usage counts.

Evidence gateevery referenced source is admitted, healthy and inside the approved tenant and dataset scope

Refusesto merge definitions — it presents the conflict to the owners and a person decides

semantic.proposal3 declared outputsconfidential

03engineering.quality.test_suite

Quality · Test Suite

Row counts, null rates, referential integrity, freshness and distribution drift per model, run on a schedule.

Evidence gateeach model declares its expectations

Refusesto mark a model verified without a test that passed

dataquality.assessment4 declared outputsrestricted

04engineering.lineage.trace

Lineage · Trace

Column-level lineage from source through models to published assets.

Evidence gateevery referenced source is admitted, healthy and inside the approved tenant and dataset scope

Refusesto present partial lineage as complete — it names where the trace stops

dataquality.assessment3 declared outputsrestricted

05engineering.schema.drift_watch

Schema · Drift Watch

Compare current source schemas against the last verified snapshot and report additions, removals and type changes.

Evidence gatea verified snapshot exists

Refusesto auto-adopt a schema change — it reports and a person decides

dataquality.issue4 declared outputsrestricted

06engineering.freshness.sla_attainment

Freshness · Sla Attainment

Compare actual arrival time per source against its declared freshness commitment, reporting attainment and the worst cases.

Evidence gateeach source declares a freshness commitment

Refusesto report attainment for a source with no declared commitment — it reports the gap instead

datasource.freshness4 declared outputsrestricted

07engineering.cost.query_spend

Cost · Query Spend

Query cost by model, dashboard and person, identifying the small number of expensive things that usually account for most of it.

Evidence gatequery logs carry a cost or scan measure

Refusesto attribute cost to a person without the asset context — a scheduled report runs under somebody's name and is not their doing

benchmark.analysis5 declared outputsinternal

08engineering.test.coverage

Test · Coverage

Coverage of declared expectations against models, weighted by how many published assets depend on each.

Evidence gateevery referenced source is admitted, healthy and inside the approved tenant and dataset scope

Refusesto report coverage as a single percentage without weighting by dependants — an untested model under 40 reports is not the same as one under none

dataquality.assessment4 declared outputsinternal

Effective capability boundary

Resolved centrally. Scoped to the approved seat.

The Role Pack cannot grant itself authority. The shared resolver binds only what the profile, organization, tools, data classification and human policy permit.

Can read · 21

analysis · assumption
analysis · comparison
analysis · confidence
analysis · lineage
benchmark · analysis
dataquality · assessment
dataquality · issue
datasource · catalogue

+ 13 more governed capabilities

Can prepare · 11

analysis · assumption
analysis · confidence
benchmark · analysis
dataquality · assessment
dataquality · issue
datasource · profile
metric · health
metric · reconciliation

+ 3 more governed capabilities

Can recommend · 8

analysis · confidence
benchmark · analysis
dataquality · assessment
dataquality · issue
datasource · freshness
metric · health
metric · reconciliation
query · result

Can request approval for · 1

semantic · proposal

Fixed boundaries

What this role will not do.

Approve own request.
Bypass approval.
Impersonate a person.
Run raw sql supplied in a prompt.
Alter a source system.
Verify own analysis.
Hide query lineage.
Change semantic model without approval.
Delete cited definition.
Mark model verified without passing quality evidence.
Resolve metric ambiguity without owner.
Change definition without stating affected count.
Adopt schema change automatically.
Resolve definition conflict unilaterally.

Connectors

Give this role only the sources its work requires.

Choose from 1,000 named connector routes. Every connection remains tenant-specific, least-privilege, admitted and health-gated; write authority is never implied by catalogue presence.

Explore all connectors All eight roles

Analytics Engineer · Data Intelligence Role Pack

Add specialist depth to an approved seat.

ServAI proposes the exact role manifest. Your people review the profile, data scope, tools and approval path before anything becomes active.

Build My AI Team Discuss a deployment

Analytics Engineer

Choose the role by the work—not by a title alone.

One responsibility for every named analysis.

8 analyses with evidence gates and refusal conditions.

Semantic · Model Change

Definition · Consolidation

Quality · Test Suite

Lineage · Trace

Schema · Drift Watch

Freshness · Sla Attainment

Cost · Query Spend

Test · Coverage

Resolved centrally. Scoped to the approved seat.

What this role will not do.

Give this role only the sources its work requires.

Add specialist depth to an approved seat.

Use this analyst depth across the Industry Packs you actually own.