Content gaps and improvement plan (2026-11-07)¶
A gap review of data.noblerworks.com, done by listing the tree with git ls-files and searching it with git grep. A topic is called missing only when the search found nothing (or one passing mention). The open items are tracked in TODO.md under "Content gaps (2026-11-07 review)".
Where the site stands¶
| Area | Has | Compared with the cloud site |
|---|---|---|
| Concept pages | 23 | Strong on modelling, pipelines and SQL |
| Day-one on-ramp | none | Cloud has 15 day-one pages |
| Topic indexes | 2 (databases, data engineering) | Cloud has 7 |
| Hands-on projects | 4 + lab map | No CDC, orchestration or lakehouse build |
| Service comparisons | 2 (databases, messaging) | Cloud has 7 |
| Roadmaps | 2 (data engineer, database specialist) | No analyst or analytics engineer path |
| Postmortems | 2 | |
| Cert guides | 24, all over 8,000 words | 6 have no flashcard deck (all 4 Snowflake, DP-203, Atlas admin) |
The site serves data engineers and DBAs well. Its gaps are the first step (there is no on-ramp for someone who has never written SQL) and the analyst side of the field.
Plan, in order¶
Wave 1 - the on-ramp¶
A learn/day-one/ section, the way the cloud site has one: first SQL query (SELECT, WHERE, GROUP BY), spreadsheets to SQL, Python and pandas or Polars for data, notebooks, CSV/JSON/Parquet in practice, and a local database (DuckDB or Postgres in Docker). Link it from learn/data-from-scratch.md.
Wave 2 - comparisons and roadmaps¶
- Service comparisons: warehouses (Redshift, Fabric/Synapse, BigQuery, Snowflake, Databricks SQL), ETL and orchestration services (Glue, Data Factory, Dataflow, managed Airflow), and BI tools (Power BI, Looker, Tableau, QuickSight, Superset). These products appear on many pages but no page compares them.
- Roadmaps: data analyst and analytics engineer (an interview guide exists for the second, but no roadmap).
Wave 3 - concept pages (the TODO candidates plus new ones)¶
- From TODO: backup and point-in-time restore; short concept pages for query plans, Spark shuffles and skew, and semantic layers (long-form versions exist).
- New: data observability (one passing mention), warehouse cost control / FinOps for data, database normalisation for OLTP, time-series and graph databases, feature stores (cross-link to ai.noblerworks.com).
Wave 4 - projects and troubleshooting¶
- Hands-on: CDC with Debezium into a lakehouse table, an Airflow (or Dagster) orchestration project, an Iceberg lakehouse on local storage, and a backup-and-restore drill.
- Troubleshooting: replication lag, deadlocks and lock waits, Kafka consumer lag.
- Topic indexes: streaming, analytics and BI, data governance.
- Postmortems: one streaming or warehouse incident with a public writeup (choose from primary sources only).
Wave 5 - certs (already in TODO)¶
- Flashcard decks for the 6 guides without one.
- New guides once codes are verified: Databricks Data Analyst and Spark Developer, Google Associate Data Practitioner, Confluent CCAC, MongoDB Data Modeler.
- Decide DP-203 (retired): keep as reference or remove.
How to do each item¶
Every new page uses the house style in CLAUDE.md. Projects must be run end to end before they ship (the dbt project is the model). Run check-readme-counts.py --fix in the same commit.