Skills
The toolkit, mapped as a network.
Data Analytics & BI is the core. Data engineering is what keeps it trustworthy, and Generative AI is what extends it. The skills in the middle — Python, SQL Server, Microsoft Fabric, star-schema modelling — are the bridges that let one person own all three.
BY DISCIPLINE
The same map, by role.
Machine learning is listed on its own rather than folded into BI — the data-science background stands separately from the reporting work, even though the two meet in every forecast.
Establish what’s true
Reporting people trust — built, published and kept running inside the business.
- Enterprise dashboard development Power BI · Tableau
- Semantic modelling star schema · relationships
- Direct Lake semantic models Fabric · OneLake
- DAX measures · time intelligence
- Calculation groups & field parameters Tabular Editor
- SQL reporting T-SQL · stored procs
- Data shaping Power Query (M)
- Composite models & aggregations import · DirectQuery
- Enterprise-scale models XMLA endpoint · large models
- Row-level & object-level security RLS · OLS
- Performance tuning DAX Studio · Tabular Editor · Performance Analyzer
- Power BI Service workspaces · apps · capacity
- Source control & CI/CD .pbip · Git · deployment pipelines
- Certified datasets & sensitivity labels endorsement · Purview
- Gateways & scheduled refresh on-prem gateway · refresh windows
- Incremental refresh partitions · large models
- Tenant automation REST API · PowerShell
- Paginated reports pixel-perfect · SSRS-style
- Requirements & KPI workshops definitions agreed up front
- User training & adoption handover · documentation
- Web & marketing analytics Google Analytics 4
- Excel modelling Power Pivot · hand-offs
Make it trustworthy
Infrastructure that turns scattered source data into a governed, queryable asset.
- Lakehouse & warehouse design Microsoft Fabric
- OneLake & shortcuts reference data in place
- Capacity management Fabric F-SKUs · monitoring
- Pipeline orchestration Azure Data Factory
- Transformation flows Dataflows Gen2
- Medallion architecture bronze · silver · gold
- ETL development ingest · clean · load
- Incremental loading watermarks · change capture
- Dimensional modelling star schema · SCD type 2
- Data quality gates tests · quarantine
- Schema drift handling log, don’t fail
- SQL Server T-SQL · staging · warehousing
- API extraction REST · paging · auth
- Lineage & auditability source row to report figure
- Version control Git · documented logic
- Freshness SLAs & alerting failures surface fast
Act on it
Production AI systems and automation that run with minimal oversight.
- LLM application development LangChain
- Agentic systems multi-step · multi-agent
- RAG pipelines chunking · embedding · retrieval
- Vector databases Supabase · Postgres
- Knowledge base automation crawl · refresh · retry
- Structured output validation schema · auto-repair
- Cost-aware model routing frontier · local
- Local model deployment Ollama
- Persistent memory Postgres-backed context
- Workflow automation n8n
- Tool & API integration REST · webhooks
- Human-in-the-loop design approval before action
- Run telemetry tokens · cost · retries
- Python engineering the glue across the stack
Say what happens next
Statistical and machine-learning work: the background behind the forecasting and scoring.
- Forecasting regression · seasonality
- Classification & scoring churn · lead · risk
- Model evaluation MSE · RMSE · R²
- scikit-learn training · pipelines
- pandas wrangling · feature prep
- NLP & sentiment text becomes a measure
Need a specific combination of these?
Most real projects pull from more than one column — that's the point of the overlap. Let's talk about what yours needs.