- Python Weekly
- Posts
- Python Weekly (Issue 736 March 12 2026)
Python Weekly (Issue 736 March 12 2026)
Rahul Chaudhary
March 12, 2026
In partnership with
Welcome to issue 736 of Python Weekly. Let's get straight to the links this week.
Ship Docs Your Team Is Actually Proud Of
Mintlify helps you create fast, beautiful docs that developers actually enjoy using. Write in markdown, sync with your repo, and deploy in minutes. Built-in components handle search, navigation, API references, and interactive examples out of the box, so you can focus on clear content instead of custom infrastructure.
Automatic versioning, analytics, and AI powered search make it easy to scale as your product grows. Your docs stay accurate automatically with AI-powered workflows with every pull request.
Whether you're a dev, technical writer, part of devrel, and beyond, Mintlify fits into the way you already work and helps your documentation keep pace with your product.
Articles, Tutorials and Talks
The post explores building a Python framework with a Rust core and the challenge of coordinating two asynchronous event loops running in both languages at the same time. It highlights the trade-off of this hybrid approach: Rust provides performance and efficiency, while Python offers developer productivity, but integrating the two runtimes adds architectural complexity.
The post presents an optimization ladder for Python performance, showing that speed improvements come from progressively moving up layers from simple code tweaks to alternative runtimes and compiled languages. It argues that Python’s dynamic design makes deep optimization difficult, so the biggest gains often come from changing the execution layer (e.g., JITs, compiled extensions, or other languages) rather than micro-optimizing Python code itself.
This study evaluates the trade-offs of Python 3.14.2's experimental "free-threaded" build, finding that while it can quadruple performance and energy efficiency for independent parallel tasks, it increases energy consumption by up to 43% for sequential code. The removal of the Global Interpreter Lock (GIL) also leads to higher memory usage and performance degradation in workloads with high object contention, indicating it is a specialized tool rather than a universal upgrade.
Philip Zucker explains "thinnings" as mathematical witness data for sublist questions, showing how they can be represented as bitvectors or prefix sums to efficiently manage de Bruijn index shifting in terms with binders. He argues that by treating thinnings as composable "mega-shifts," developers can create more efficient, auditable workflows for lambda e-graphs, hash consing, and variable management without the complexity of dependent type theory.
The tutorial teaches DSPy by building progressively more complex deep research agents that search the web, gather sources, and synthesize reports. Along the way it explains core DSPy concepts such as signatures, modules, and composable AI programs, showing how developers can build reliable agent systems without heavy prompt engineering.
Python’s logging system goes far beyond basic message output. This guide explores handlers, structured logging, performance trade-offs, and real-world patterns for building production-ready logging pipelines.
Nine Python agent frameworks, compared honestly. Architecture, code samples, community sentiment, and what actually matters when you're picking one.
The article introduces Madblog, a lightweight blogging engine that lets you publish posts directly from a folder of Markdown files without databases, plugins, or complex CMS infrastructure. It adds built-in Fediverse federation (ActivityPub) and Webmentions, allowing blog posts to be followed, replied to, and shared across platforms like Mastodon while keeping content fully self-hosted.
This post introduces ascend, an open-source Python library designed to execute functions on Kubernetes with minimal code changes and no local container requirements. It leverages cloudpickle for serialization and communicates directly with the Kubernetes API to manage jobs, scaling infrastructure to zero when not in use.
The video demonstrates refactoring a FastAPI + SQLAlchemy application into a Ports and Adapters (Hexagonal Architecture) design that isolates domain logic from frameworks. By introducing domain types, errors, protocols as ports, and adapters for infrastructure, the result is clean, framework-independent business logic that is easier to test, maintain, and evolve.
ChronoName is a Python-based utility that uses ExifTool to rename photos and videos deterministically based on their true capture metadata, resolving timezone discrepancies between still images and QuickTime videos. It prioritizes safety with features like dry-run modes, undo logs, and an auditable manifest to ensure large media archives remain organized and reversible at the filesystem level.
FinePhrase is a newly released (March 2025) Hugging Face Space and dataset by the Hugging Face Science Team (FineData) that serves as an interactive "Synthetic Data Playbook" for generating trillions of high-quality pre-training tokens. It demonstrates a scalable pipeline for adapting pre-training data processing to over 1,000 languages, moving beyond traditional web-crawled HTML to leverage synthetic generation for filling gaps in low-resource and high-quality educational content.
Books
This is a practical guide to help you transform from Machine Learning novice to skilled Machine Learning practitioner. Throughout the book, you’ll learn the best practices for proper Machine Learning and how to apply those practices to your own Machine Learning problems.
Interesting Projects, Tools, and Libraries
AI agents running research on single-GPU nanochat training automatically.
Self-hosted platform to unify wearable health data through one AI-ready API.
A lightweight pandas-compatible DataFrame API built on top of Polars for speed and simplicity.
django-paradedb connects ParadeDB’s pg_search PostgreSQL extension with the Django ORM to provide Elasticsearch-quality search using BM25 indexes. It integrates directly with Django models and querysets, enabling powerful search queries without leaving the ORM.
OpenViking is an open-source context database designed specifically for AI Agents(such as openclaw). OpenViking unifies the management of context (memory, resources, and skills) that Agents need through a file system paradigm, enabling hierarchical context delivery and self-evolving.
LLM inference, optimized for your Mac. Continuous batching and tiered KV caching, managed directly from your menu bar.
OpenRAG is a comprehensive, single package Retrieval-Augmented Generation platform built on Langflow, Docling, and Opensearch.
Opinionated, off-the-shelf observability for Django. Drop it in and get traces, structured logs, metrics, and profiling wired up with sensible defaults.
Your personal finance dashboard, fully automated - local-first, export-based.
A lightweight sprint dashboard for Gitea. FastAPI + HTMX + SQLite — no JavaScript framework, no build step.
An open-source AI tool for high-quality green-screen keying, using neural models to separate foreground and background while preserving fine details like hair, motion blur, and transparency.
reverse engineering Gemini's SynthID detection.
model-native and tensor-optimized CPU research runtimes with organized workloads, tools, and docs
Upcoming Events and Webinars
There will be following talks
Applications with Azure AI Foundry
Astronomers have too many images. What they really want is time series
There will be a talk, Investing for financial freedom with data and statistical reasoning
There will be a talk, Information Theory in the Age of AI.
There will be a talk, AI: Automating Your Workflow with Python & LLMs.
There will be a hands on tutorial, Python Ops: Jinja + Ansible.
There will be following talks
Data & AI at Elisa
Optimization problems in electricity markets
How fast can you read data from Delta Lake without a cluster?
Goodbye argparse, Hello Typer
Teaching AI to Write Poetry in Three Languages: Fine-Tuning mBART for a Museum Exhibit
Our Other Newsletters |
Programmer Weekly - A free weekly newsletter for programmers.
Founder Weekly - A free weekly newsletter for entrepreneurs featuring best curated content, must read articles, how to guides, tips and tricks, resources, events and more.