Architectural guide
Medallion architecture for engineers: building trusted data products
Medallion architecture for engineers: design raw data, trusted views and published data products with clear contracts, replay paths and ownership.
Beyond the Notebook
Maciej Kępa
Data & AI Architect
I write about building data and AI systems that survive production — MLOps, observability, platform engineering and architecture.
Current focus
Production over demos
Practical notes on operating AI systems under real production constraints.
Platforms
Azure Databricks
Topics
AI Data MLOps
Public work
Articles Conference talks
Featured writing 
Long-form notes for engineers who have to ship
Architecture essays, delivery guides and operational lessons from the part of AI work that starts after the first promising notebook.
Architectural guide
Feature store alternatives: feature tables, transformation libraries and warehouse-first patterns
Feature store alternatives for production ML: when feature tables, shared transformation libraries or warehouse-first patterns are enough, and when a feature store earns its cost.
Architectural guide
ML release management: what a model release needs beyond the registry
ML release management needs more than a registered model. Learn how MLflow supports versioning, validation and rollback on Azure Databricks.
Topics
Choose a production AI reading path
12 posts
Production MLOps
From model packaging to deployment, ownership and operational readiness.
Explore MLOps → 7 postsDatabricks
Lakehouse-side MLOps, delivery patterns and production data platform decisions.
Explore Databricks → 3 postsML Observability
Monitoring signals, feedback loops and operational baselines.
Explore observability → Speaking
Selected talks
Conference sessions on production AI, MLOps, observability and data platform architecture.
SQL Server and the Vector Revolution - how LLMs see your data
Embeddings, vector search and the bridge between traditional data platforms and modern AI.
BigQuery, Vertex AI or Databricks? Don’t choose - combine them!
A reference architecture for building one AI platform across BigQuery, Databricks and Vertex AI without creating new silos.
About me
Architecture, delivery and AI systems that have to hold up in production
I am a Data & AI Architect working on production data platforms, MLOps and cloud-native AI systems. My work focuses on the engineering decisions that make these systems operable, observable and maintainable.