Cheat Sheet Phase 6 : Specialisation & Leadership

2. CDMP - 14 Domaines DMBOK

Niveaux : Associate ≥ 400, Practitioner ≥ 500, Master ≥ 600 (sur 700)

3. Snowflake SnowPro Core

Pieges courants

• Time Travel : 1 jour (Standard), 90 jours (Enterprise)
• Clustering : seulement tables > 1TB, pas de cluster key sur colonnes haute cardinalite + filtrage rare
• Warehouses : auto-suspend min 1 min, scaling policy Standard vs Economy
• Zero-copy clone : ne clone PAS les privileges

4. GCP Prof. Data Engineer

Services cles a maitriser

5. Databricks & AWS Certifications

Databricks DE Associate - Domaines

• Databricks Lakehouse Platform (24%)
• ELT with Spark SQL & Python (29%)
• Incremental Data Processing (22%)
• Production Pipelines (25%)

AWS Data Engineer Associate - Domaines

• Data Ingestion & Transformation (34%)
• Data Store Management (26%)
• Data Operations & Support (22%)
• Data Security & Governance (18%)

6. Confluent & dbt Certifications

Confluent Certified Developer

• Kafka Fundamentals (15%)
• Kafka Architecture (20%)
• Producers (20%)
• Consumers (20%)
• Kafka Streams & ksqlDB (15%)
• Connect & Schema Registry (10%)

dbt Analytics Engineering

• dbt Models & Materializations
• Tests & Documentation
• Jinja & Macros
• Packages & Seeds
• Snapshots & Incremental

dbt est gratuit et peut etre passe en ligne a votre rythme.

7. Strategies de Preparation

Plan type (6 semaines)

Techniques efficaces

• Spaced Repetition : Anki avec flashcards personnalisees
• Active Recall : fermer le livre, ecrire de memoire
• Pomodoro : 25 min focus + 5 min pause
• Feynman : expliquer a un debutant pour verifier la comprehension
• Practice tests : minimum 3 mock exams complets avant le vrai

8. Portfolio Data Architect

Checklist du portfolio ideal

Structure README type

# Project Name
## Architecture Overview (C4 diagram)
## Tech Stack & Justification
## Setup & Installation
## Data Flow Description
## Key Design Decisions (ADR links)
## Results & Metrics
## Lessons Learned

10. Real-time Pipeline - Stack

Composants

11. Data Mesh - Implementation

4 Principes fondamentaux

• Domain Ownership : chaque domaine possede ses donnees
• Data as a Product : SLAs, qualite, documentation
• Self-serve Platform : infrastructure en libre-service
• Federated Governance : standards globaux, autonomie locale

Centralise vs Mesh

12. Contribution Open Source

Workflow de contribution

1. Fork le repo sur GitHub
2. git clone + branch feature
3. Code + tests + docs
4. git push + Pull Request
5. Code review + iterations
6. Merge (maintainer)

Ou contribuer (Data)

• dbt : packages (dbt-utils, dbt-expectations), adapters
• Airflow : providers, operators, hooks
• Great Expectations : custom expectations
• Airbyte : connecteurs source/destination
• Apache Iceberg : docs, bug fixes

13. C4 Model - Documentation Architecture

Outils

• Structurizr : DSL + rendu automatique (recommande)
• PlantUML : C4-PlantUML extension
• Mermaid : integre dans GitHub Markdown
• draw.io : drag-and-drop, shapes AWS/GCP/Azure

Structurizr DSL

workspace {
  model {
    user = person "Data Analyst"
    platform = softwareSystem "Data Platform" {
      ingestion = container "Airbyte"
      warehouse = container "Snowflake"
      transform = container "dbt"
      bi = container "Metabase"
    }
    user -> bi "Consulte dashboards"
    bi -> warehouse "SQL queries"
    transform -> warehouse "Transformations"
    ingestion -> warehouse "Raw data load"
  }
  views {
    systemContext platform "Context" {
      include *
      autoLayout
    }
    container platform "Containers" {
      include *
      autoLayout
    }
  }
}

14. ADR - Architecture Decision Record

# ADR-NNN : [Titre de la decision]

## Status
[Proposed | Accepted | Deprecated | Superseded]

## Date
YYYY-MM-DD

## Context
Probleme business et technique a resoudre.
Contraintes : budget, timeline, equipe, legacy.

## Decision
La decision prise avec justification.

## Alternatives considerees
| Option | Avantages | Inconvenients | Rejetee car |
|--------|-----------|---------------|-------------|
| Option A | ... | ... | ... |
| Option B | ... | ... | ... |

## Consequences
### Positives
+ [Avantage 1]
+ [Avantage 2]
### Negatives
- [Inconvenient 1]
- [Risque mitige par ...]

## Related ADRs
- ADR-XXX (depend de)
- ADR-YYY (remplace)

15. Stakeholder Management

Matrice Pouvoir / Interet

Communication par audience

16. ROI Data Platform - Formules

Presentation COMEX - Template 3 slides

• Slide 1 - Le Probleme : "Nous perdons X EUR/an parce que..." (business, pas technique)
• Slide 2 - La Solution : 1 diagramme C4 Level 1, 3 bullet points max
• Slide 3 - Le ROI : Investissement X EUR, retour Y EUR/an, payback Z mois

18. Vendor Evaluation - Scoring Matrix

Build vs Buy

19. Data Strategy Roadmap - Template

Structure en 4 phases

Frameworks de priorisation

• RICE : Reach x Impact x Confidence / Effort
• MoSCoW : Must / Should / Could / Won't
• Value vs Effort : matrice 2x2 (quick wins en premier)

20. RACI & Checklist Leadership

Matrice RACI Data Platform

R = Responsible, A = Accountable, C = Consulted, I = Informed

Checklist Data Architect Leader

• ADRs a jour pour chaque decision majeure
• Architecture Review Board mensuel
• Data quality dashboard visible par tous
• Roadmap data mise a jour trimestriellement
• 1:1 hebdo avec chaque tech lead
• Blog interne / newsletter data mensuel
• Skills matrix de l'equipe a jour
• Budget FinOps revise mensuellement