Blog
Practical insights, opinions and stories from the Beez about Python, cloud, data and everything in between.
uv as a poetry/pyenv/pipx replacement
How the recently expanded uv consolidates poetry, pyenv and pipx into a single fast Python toolchain.
K8S Security
Why Docker and Kubernetes are not secure by default, and how to harden UID mapping, image layers, RBAC, network policies and secrets for FastAPI.
The memory footprint of your Python application
A tour of Python's memory model: how objects are allocated, stored and cleaned up, with practical tweaks to cut memory usage.
Vector databasing with DuckDB on top of PGVector
An architecture integrating DuckDB with Postgres and PGVector for vector databasing, bridging the gap between OLTP and OLAP.
Streamlining Azure Logic Apps Deployment with Managed Identity in Azure DevOps: An IaC Approach
How to deploy Azure Logic Apps securely with Managed Identities, Bicep and Azure DevOps YAML pipelines as code.
So you want to expose a database? An exploration of automated API generation
Hackathon write-up comparing automatic and AI-driven ways to generate a CRUD API from an existing PostgreSQL database.
EuroPython 2023 Trip Report
A first-hand report from EuroPython 2023 in Prague: keynotes, performance tricks, AI talks and key takeaways.
Type hinting in modern Python: The Protocol class
How Python's Protocol class brings static type checking to duck-typed code without forcing explicit inheritance.
A reproducible, terminal-first Python development box
How to build a reproducible, keyboard-first Python development box with Ansible, a terminal-based workflow and the right tooling.
Secure MLOps with Databricks MLFlow
How to run multiple logical environments in one Databricks workspace and securely separate production models with an environment-aware MLFlow client.
Build Stuff 2022: Summaries
Notes and summaries from the most valuable talks at the Build Stuff 2022 conference in Vilnius.
Blipz on the radar 2022: summaries
Summaries of the Blipz 2022 talks: data mesh, sustainable IT, Polars, CDK, human-in-the-loop ML, OSINT and zero trust.
EuroPython 2022: Summaries of selected talks
Codebeez summaries of selected EuroPython 2022 talks, covering the JWST pipeline, error messages, PyArrow, asyncio debugging and clean architecture.
The magical fusion between batch and streaming insights
How the Lambda architecture fuses batch and streaming insights into fresh data products, with our Azure implementation as illustration.
Elegant CICD with Databricks notebooks
How to unit and integration test Databricks notebooks on transient clusters and ship them as reproducible artifacts via Azure DevOps.
Single-node and distributed Deep Learning on Databricks
How single-node and multi-node Databricks clusters enable parallel and distributed Deep Learning beyond Spark, using Data Factory, Horovod and Petastorm.
The Netherlands, a worldwide AI knowledge hub
How the Netherlands could become a worldwide AI knowledge hub by leading in the ethical, societal application of high tech.
MLOps: a tale of two Azure pipelines
How to combine Azure DevOps and AzureML pipelines into an end-to-end MLOps solution for continuous training and deployment.
AzureML PyTorch GPU enabled compute target: Unifying remote and local environments
Configure an AzureML GPU compute target for PyTorch and reuse the same Docker image to unify local development and remote training.
Microsoft and Python Machine Learning, a modern love story, Part 1
A development workflow for launching ML models in the cloud with AzureML while coding in VSCode devcontainers on WSL2.