Skip to content

Coverage v3 — A DEA-C01 Subset

HungerSync Case Series — Adding a DEA-C01 Subset (Coverage v3)

Section titled “HungerSync Case Series — Adding a DEA-C01 Subset (Coverage v3)”

Extends coverage-v2-mla-c01.md. Verdict: include a surgical subset, not the whole cert. DEA’s distinctive core (Data Store Management) is the one layer none of the three current certs touch — the lakehouse/warehouse beneath the prediction engine. Most of the rest of DEA overlaps MLA-C01 / AIP-C01 / existing cases and is absorbed via tags, not duplicated. This subset completes the stack bottom-to-top; treat it as the floor.


LayerCertHungerSync surface
0 · Data platformDEA-C01 (subset)lakehouse: ingest → store → catalog → model (star schema) → serve/analyze
1 · PredictionMLA-C01delay/demand/taste models, trained on the platform
2 · ApplicationAIP-C01GenAI app: RAG, FM integration, safety, ops
3 · AgentClaude Architectagentic loops, MCP, Claude Code, dispatch

Reads upward: the lakehouse feeds the models; the models feed the app; the app drives the agent. No layer exists below data or above agent — the taxonomy is now complete.

2. DEA-C01 at a glance (pass 720/1000; 50 scored Qs)

Section titled “2. DEA-C01 at a glance (pass 720/1000; 50 scored Qs)”
DomainWeightIn the subset?
D1 Data Ingestion & Transformation34%Partial (1.1 depth, 1.3)
D2 Data Store Management26%Yes — the core
D3 Data Operations & Support22%Partial (3.1, 3.2, 3.4)
D4 Data Security & Governance18%Absorbed (tags on CS8/CS9)

INCLUDE (net-new — the data-platform layer → new cases DE1/DE2):

DEA taskWhereWhy net-new
1.1 Perform data ingestion (depth: replayability, fan-in/out, throttling, event triggers, schedulers)DE1platform-grade ingestion beyond MLA’s feature-store ingest
1.3 Orchestrate data pipelines (MWAA, Step Functions, Glue workflows, fault tolerance)DE2pipeline orchestration as a discipline
2.1 Choose a data store (Redshift, EMR, Lake Formation, Iceberg, vector index types)DE1untouched by other certs
2.2 Data cataloging (Glue Data Catalog, crawlers, partition sync)DE1untouched
2.3 Data lifecycle (S3 tiering, TTL, versioning, retention, deletion)DE1untouched
2.4 Data models & schema evolution (star schema, partitioning, lineage, compression)DE1your day-job core; untouched
3.1 Automate data processing (Athena, Glue, EMR, EventBridge)DE2serving/automation
3.2 Analyze data (Athena, QuickSight, SQL views, aggregation/pivot)DE2analytics/BI layer (the “Revenue-by-concourse” surface)
3.4 Ensure data quality (DQ rules, consistency, sampling, skew)DE1platform DQ discipline

ABSORB (overlaps existing coverage — add DEA tags, do NOT duplicate):

DEA taskAbsorbed into
1.2 Transform & processML1 + AIP 1.3 (note 1.2.10 “LLMs for data processing” → tag CS4/CS5)
1.4 Programming concepts (IaC, CI/CD, distributed)Claude D3, MLA 3.2/3.3, DE2 (light)
3.3 Maintain & monitor pipelinesCS9 (ops center)
4.1 Authentication (VPC, IAM, Secrets Manager)CS8
4.2 Authorization (IAM, Lake Formation, RBAC/TBAC/ABAC)CS8
4.3 Encryption & masking (KMS)CS8
4.4 Audit logs (CloudTrail, CloudWatch, CloudTrail Lake)CS8 / CS9
4.5 Privacy & governance (Macie PII, residency, data sharing, sovereignty)CS8

CaseTitleLayerDEA tasksBoundary
DE1The lakehouse: storing & modeling HungerSync’s data01.1, 2.1, 2.2, 2.3, 2.4, 3.4The general-purpose lakehouse + star schema that all consumers read.
DE2Pipelines & analytics: orchestrating and serving01.3, 3.1, 3.2Orchestration + BI/analytics over the star schema (concession revenue by concourse, demand by terminal).

Critical boundary vs ML1 (avoids ingestion overlap): DE1 is the platform — the lakehouse and dimensional model that serves analytics, ML, and the app alike. ML1 (MLA-C01) is the ML-specific feature engineering and training-data prep that consumes from DE1’s lakehouse into the Feature Store. This mirrors the real-world split between a data-platform team and an ML team and keeps the two cases distinct.

DE1 is essential; DE2 is optional — if you want it tighter, fold analytics into DE1 and accept lighter coverage of DEA 3.2.


  • Strong: DEA D2 (26%, all four tasks) via DE1 — the cert’s distinctive core.
  • Good: D1.1/1.3 and D3.1/3.2/3.4 via DE1/DE2.
  • Incidental only (absorb-tags): D1.2, D1.4, D3.3, and all of D4 (18%).

So this subset makes the data-platform layer of the world real and teachable, and covers ~70%+ of DEA by weight. It does not by itself make you DEA-exam-ready — full readiness would need dedicated drilling on D4 (security/governance) and the absorbed ops/transform items, which the case series touches but does not drill. That’s the intended tradeoff of “subset.”


6. Updated full lineup (14 cases + capstone)

Section titled “6. Updated full lineup (14 cases + capstone)”

C=Claude · A=AIP-C01 · M=MLA-C01 · D=DEA-subset

LayerCases
0 Data platformDE1 (D), DE2 (D)
1 PredictionML1 (M), ML2 (M), ML3 (M/A)
2–3 App & AgentCS1 (A/M/C), CS2 (C/A), CS3 (C+A), CS4 (A/C), CS5 (C/A), CS6 (C/M), CS7 (C/A/M), CS8 (A/M/D), CS9 (A+C/M/D)
CapstoneCS10 (all four layers)

Tag additions from this subset: CS8 gains D (DEA D4 governance/security depth), CS9 gains D (DEA D3.3 pipeline monitoring), CS10 capstone now spans four layers.


This subset is the floor of the stack. The four layers (data → models → app → agent) form a complete vertical; there is nothing below data or above agent in this frame. Recommendation: stop adding certs here. Further additions (e.g., a networking or pure-Redshift cert) would add overlap, not a new layer, and would dilute the world’s coherence. If a genuinely new layer ever appears (e.g., a hardware/edge robotics certification for full robot-fleet autonomy), that would justify reopening — nothing in the current candidate set does.


  1. DE2 in or folded? (Lean: keep DE1; DE2 optional. DE1 alone covers the distinctive D2 core; DE2 adds the analytics/orchestration breadth.)
  2. Confirm CS10 capstone now spans all four layers.
  3. Build order unchanged: template → CS1 (reference). DE1 is the natural bottom of the stack and a strong second reference case given it’s closest to your real work.