Maturity Model · 03 of 3

DataOps & AI

Six DataOps dimensions and three levels of AI-assisted engineering.

StrategyArchitectureAutomationQualityGovernanceSkillsSix dimensions
01

DataOps: Six Dimensions

NGT's DataOps assessment scores six dimensions into one scorecard.

02

From Scorecard to Running Platform

  1. 01Assess
    • Score the six dimensions
    Scorecard, gaps, roadmapGo / no-go
  2. 02Prioritise
    • Use cases with the clearest pain
    Pilot scopeGo / no-go
  3. 03Pilot
    • Prove it on one data product
    Proven patternGo / no-go
  4. 04Scale
    • The standard platform for more teams
    Shared platform
  5. 05Run
    • SLAs, pipeline monitoring, runbooks
    Platform as a service
03

AI-Assisted Engineering: Three Levels

FPT Software Hanoi's framework for AI in the software lifecycle. What changes is how approval works.

  1. 01L3 · AI-orchestratedAgentic, outcome-driven work. Approval is policy-gated, with audit and drift checks.Measure: AI code surviving in production
  2. 02L2 · AI-integratedMeasured and built into the workflow. AI proposes, a human approves.Measure: accepted completionsPR resolution time
  3. 03L1 · AI-assistedAd hoc and human-led; AI is a sidekick. The engineer decides every time.Measure: baseline velocityDefect density
A person stays accountable at every level

Questions on Scoring

The questions this part raises, and their answers.

No. The target is agreed per area; some reach their goal at Challenger.

SourcesNGT Cloud, DevOps and AI Service Offerings v2.0 (Mar 2025)NGT DevSecOps Service Offering (Jan 2025)Puppet, State of DevOps research (five stages of DevOps evolution)NGT Platform Engineering v1.2 (Sep 2026)NGT DataOps Service Offering drafts (Sep 2026)FPT Software Hanoi AI Service Introduction v0.9.3 (Mar 2026)