Course Objective & Scope
When 'net revenue' or 'active customer' is calculated with slightly different SQL across dashboards, leadership loses trust in data—and Text-to-SQL or LLM agents querying raw tables directly hallucinate joins and filters. This training teaches how to build a modern semantic layer that defines metrics, entities and relationships once as code.
dbt Semantic Layer (MetricFlow) and Cube architectures, defining entities, dimensions and measures, ratio and cumulative metrics, pre-aggregations, and serving governed metrics to both BI tools and LLM agents via SQL, REST and MCP interfaces are covered hands-on. Participants produce a semantic layer blueprint for their organisation's core KPIs.
Who Should Attend?
- Analytics engineers and BI architects responsible for metric standardisation
- AI and data engineers building Text-to-SQL systems and analytical copilot agents
- Analytics leads governing enterprise KPI definitions
- Prerequisite: SQL, dimensional modeling concepts and basic YAML/dbt familiarity
Key Takeaways
- Model business entities, dimensions and measures declaratively in a semantic layer
- Author simple, derived, ratio, conversion and cumulative metrics in one governed repository
- Prevent fan-out and chasm-trap errors across multi-fact joins using semantic join graphs
- Ground LLM and analytical agents on semantic layer APIs and MCP tools instead of raw, unconstrained SQL
- Configure pre-aggregations, caching and row-level security (RLS) for high concurrency and governance
Curriculum
Why a Semantic Layer? BI Drift and Text-to-SQL Hallucinations
- How metric logic trapped inside BI workbooks leads to conflicting numbers
- Why LLMs fail when generating raw SQL against complex warehouse schemas
Semantic Modeling: Entities, Dimensions, Measures and Metrics
- Authoring semantic models with dbt Semantic Layer (MetricFlow) and Cube
- Defining simple, ratio, derived, cumulative and conversion metrics
Time Spines, Multi-Hop Joins and Preventing Fan-Out
- Metric time spines, multi-grain time dimensions and period-over-period analysis
- Preventing chasm and fan traps automatically via the semantic join graph
Integrating with Text-to-SQL and AI Agents
- Replacing unconstrained Text-to-SQL with Text-to-Semantic-Query and MCP tools
- Enriching metric descriptions, synonyms and golden examples to boost agent accuracy
Performance, Pre-Aggregations and Access Control
- Reducing latency and warehouse cost with pre-aggregations and caching
- Centralised row- and column-level security (RBAC/RLS) and usage auditing
Other Courses in This Category
-
Metadata engineering to boost retrieval precision and security in RAG and agent systems: schema design, automated extraction, pre-filtering, temporal validity and ACL permission filtering.
- Live Course
- 5 modules
-
Web Scraping
Hands-on Web Scraping training covering scalable data extraction from static and dynamic web sources using BeautifulSoup, Scrapy, and Playwright, bypassing anti-bot defenses, LLM-powered data extraction, and data validation with Pandas.
- Live Course
- 13 modules
-
Data Science
End-to-end Data Science training spanning data analysis and machine learning with Python, Pandas, SQL, and Scikit-learn to deep learning with PyTorch/TensorFlow, LLMs, Apache Spark, and MLOps workflows.
- Live Course
- 18 modules
-
Data science lifecycle (CRISP-DM), exploratory data analysis, core machine learning concepts and model evaluation: a first end-to-end project for analysts and business teams who want to make data-driven decisions.
- Live Course
- 5 modules
-
Data loading, cleaning, transformation and visualisation with NumPy, pandas, Matplotlib/Seaborn and scikit-learn, for analysts and developers who want to write reusable analysis notebooks in Jupyter.
- Live Course
- 5 modules
-
Descriptive statistics, probability distributions, hypothesis testing, confidence intervals, A/B testing and regression, applied in Python so data scientists and analysts can draw statistically sound conclusions.
- Live Course
- 5 modules
Get in touch
