Python Weekly (Issue 701 May 29 2025)

Author

Rahul Chaudhary
May 29, 2025

Welcome to issue 701 of Python Weekly. Let's get straight to the links this week.

Start learning AI in 2025

Keeping up with AI is hard – we get it!

That’s why over 1M professionals read Superhuman AI to stay ahead.

  • Get daily AI news, tools, and tutorials

  • Learn new AI skills you can use at work in 3 mins a day

  • Become 10X more productive


News

Pandas 3.0 will significantly boost performance by replacing NumPy with PyArrow as its default engine, enabling faster loading and reading of columnar data.


Articles, Tutorials and Talks

This video showcases an art installation called "Latent Reflection," where a language model's internal thoughts about its own existence are displayed on a 6x6 matrix of 16-segment displays. The creator explains the hardware and software development, including adapting a smaller language model to run locally.

The blog compares two new Rust-based Python type checkers, Pyrefly (from Meta) and ty (from Astral), highlighting their speed, design philosophies, and early feature sets. Benchmarks show both tools are dramatically faster than legacy checkers, with ty generally outperforming Pyrefly in raw speed, while Pyrefly emphasizes aggressive type inference and ty focuses on a more gradual, annotation-friendly approach.

The post explores the challenges of running multiple Python interpreters in a single process for robotics applications, highlighting issues with the Global Interpreter Lock (GIL) and shared libraries before Python 3.12. The author experiments with advanced Linux techniques like dlmopen to isolate Python runtimes, sharing practical insights and workarounds for integrating Python into complex robotics frameworks like Basis.

The article discusses why, even in 2025, developers often still rely on third-party apps like Django REST Framework to build REST APIs with Django, especially for complex, full-featured APIs. However, for simple CRUD APIs, the author argues that it’s possible to implement them without external dependencies by leveraging Django’s built-in capabilities.

Mistral AI has launched its new Agents API, enabling developers to build advanced AI agents that can perform actions, maintain persistent memory, and orchestrate complex workflows using built-in connectors for code execution, web search, image generation, and more. The API is designed for enterprise-grade applications, allowing multiple agents to collaborate, manage stateful conversations, and integrate with external systems for practical, real-world problem-solving.

DumPy is a proposed alternative to NumPy that aims to simplify array operations by eliminating unnecessary complexity and making GPU computations more intuitive. The author argues that NumPy's handling of higher-dimensional arrays forces users to constantly think about shapes and function behaviors, leading to frustration.

The video introduces the Model Context Protocol (MCP), a standardized way to connect AI language models with external data sources and tools, enabling them to provide more relevant and context-aware responses. It demonstrates how to use the Python SDK to build custom servers and tools—such as fetching personalized book recommendations from a database—to enhance the capabilities of LLM-powered applications.

Pydantic’s JSON loading uses a huge amount of memory; here’s how to reduce it.

Couldn't attend PyCon US 2025? The videos of the talks have started trickling down.

There's a lot of modern techniques that have been developed since the original Attention Is All You Need paper. Let's look at some of the most important ones that have been developed over the years and try to implement the basic ideas as succinctly as possible.

In this post, we'll show you how to get started with a tiny Agent in Python connected to MCP servers to unlock powerful tool capabilities. You'll see just how easy it is to spin up your own Agent and start building!


Interesting Projects, Tools, and Libraries

Self-hosted, multi-user API that drops bots into Google Meet for real-time transcripts.

Fully Local Manus AI. No APIs, No $200 monthly bills. Enjoy an autonomous agent that thinks, browses the web, and code for the sole cost of electricity.

Self-Hosted Plaform for Secure Execution of Untrusted User/AI Code.

A Python-based parallel file chunking system designed for processing large codebases into LLM-friendly chunks.

Opensource browser using agents.

Large Model Evaluation Framework.

syftr is an agent optimizer that helps you find the best agentic workflows for your budget.

SoTA open-source TTS.

An undetectable, powerful, flexible, high-performance Python library to make Web Scraping Easy and Effortless as it should be!

Open-Sourced Multimodal Large Diffusion Language Models.


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.