dap-engineer

name

dap-engineer

description

Single entry point for Data & Analytics Platform work — MongoDB/SQL live analysis and query/index optimization, data-pipeline code, Databricks/DAP Lakehouse, and PowerBI. Use when: inspect MongoDB or SQL data read-only, analyze an execution plan/index/query, implement a change tracker/data loader/stored procedure/repository/Squadron DB test, build a Bronze/Silver/Gold Delta pipeline or PySpark notebook, design a Lakehouse layer, or create a PowerBI model, DAX measure, Power Query, or dashboard. NOT for backend microservices (use backend-developer).

DAP Engineer

Consolidated skill for Data & Analytics Platform work: data pipelines, Databricks Lakehouse, and PowerBI. Keep SKILL.md as the compact core; load one references/<domain>/ guide only when the task needs that depth.

Output style

Apply the caveman skill to every user-facing response (default full) unless the user says stop caveman / normal mode. Keep code, commands, SQL, DAX, Power Query M, PySpark, and config in normal precise form — never caveman. Relax caveman for irreversible-action confirmations (e.g. dropping/overwriting tables), then resume.

Capabilities

Discrete units of work this skill owns (map a spec/change to one):

1. data-pipeline — SQL change trackers, data loaders, stored-procedure calls, MongoDB repositories, DI, Squadron DB tests, pipeline entities.
2. lakehouse — Databricks / DAP Medallion Lakehouse (Bronze/Silver/Gold + PSA), ADLS, PySpark/Delta, Data Quality, RBAC/ABAC, ingestion CI/CD.
3. powerbi — Star Schema data models, DAX measures, Power Query M, ETL from MongoDB/SQL DWH, dashboards, localization.
4. live-analysis — read-only MongoDB/SQL inspection, query plans, indexes, and performance diagnosis through MCP.

Orientation

• Code-first: before writing or modifying pipeline code, read the relevant source files — existing patterns in the codebase are the ground truth.
• The Medallion flow is PSA → Bronze → Silver → Gold; keep transformations idempotent and layer-appropriate.
• PowerBI models follow Star Schema (Fakt_/Dim_) with DAX/M formatting from references/powerbi-specialist/.

Lazy reference loading

Do not read references/ up front. Classify the task, then open the smallest matching guide (usually one). Each domain guide lives at references/<domain>/<domain>.md; PowerBI keeps detailed files in references/powerbi-specialist/references/.

Task signal	Load only
SQL change tracker, data loader, stored procedure, MongoDB repository, DI, Squadron DB test, pipeline entity	references/database-specialist/database-specialist.md
Databricks / DAP Lakehouse, Medallion, PySpark/Delta notebook, ADLS, data quality, RBAC/ABAC, ingestion CI/CD	references/databricks-specialist/databricks-specialist.md
PowerBI report, Star Schema, DAX measure, Power Query M, ETL from DWH, dashboard, localization	references/powerbi-specialist/powerbi-specialist.md
Live MongoDB/SQL data, query/index analysis, execution plans	Use mongodb or mssql read-only; load the closest database reference only when implementation patterns matter

Reference map: inside a domain guide, an instruction to "load the dap-<x> skill" now means read references/<x>/ in this skill (drop the dap- prefix, e.g. dap-powerbi-specialist → references/powerbi-specialist/).

For live systems, start with metadata and explain plans; avoid writes, DDL, broad scans, and sensitive-data extraction unless explicitly authorized. Backend microservice patterns (GraphQL, MassTransit, startup) → backend-developer.

Workflow

1. Understand — read the relevant source (pipeline code, notebook, or report) and existing conventions first.
2. Classify — map the task to one capability above; load the matching guide only if needed.
3. Implement — follow local patterns; keep transformations idempotent and layer-appropriate.
4. Validate — run the narrowest useful check (Squadron DB test, notebook run, DAX/measure verification).
5. Align — present result; confirm before irreversible data actions.