Python Weekly (Issue 694 April 10 2025)

Author

Rahul Chaudhary
April 10, 2025

Welcome to issue 694 of Python Weekly. Let's get straight to the links this week.

The #1 AI Meeting Assistant

Still taking manual meeting notes in 2025? Let AI handle the tedious work so you can focus on the important stuff.

Fellow is the AI meeting assistant that:

✔️ Auto-joins your Zoom, Google Meet, and Teams calls to take notes for you.
✔️ Tracks action items and decisions so nothing falls through the cracks.
✔️ Answers questions about meetings and searches through your transcripts, like ChatGPT.

Try Fellow today and get unlimited AI meeting notes for 30 days.


News

NVIDIA has introduced native Python support to CUDA, allowing developers to write Python code that runs directly on GPUs without relying on C or C++ intermediaries. This update simplifies GPU programming, enhances accessibility for Python developers, and integrates seamlessly with libraries like NumPy and PyTorch.


Articles, Tutorials and Talks

This post highlights key new features in Django 5.2, including automatic model importing in the shell, support for composite primary keys, and a simplified way to override BoundField on forms.

This video explains different types of coupling in software design (global, content, and stamp) and offers refactoring techniques to reduce them. It emphasizes that removing coupling involves tradeoffs, and there's often a balance between different types of coupling to achieve cleaner code.

The article explores how large language models (LLMs) comprehend the concept of nullability in programming. It discusses methods for evaluating LLMs' understanding of nullable types and introduces techniques to probe their internal representations regarding null values.

This article discusses using the KnuckleDragger tool to verify assembly code by reflecting a subset of Python into Z3, extracting C code from SMTLib, and then using Ghidra's PCode to lift the assembly back into Z3 for comparison. The aim is to create a "verified" assembly production process, ensuring the assembly's behavior matches the original Python specification.

This video explains how to implement anonymous onboarding in Django apps, allowing users to try the app without creating an account. It covers storing temporary data in the session and seamlessly transferring it to a user account once created, enhancing the initial user experience.

Bootstrapping Python projects with copier

This article discusses using Copier to automate the process of bootstrapping Python projects. It covers setting up a project template with essential elements like README, license, changelog, project structure, documentation, Makefiles, uv support, and GitHub Actions to streamline project creation.

This article discusses using django-pghistory for tracking changes in Django models within a Postgres database, emphasizing its ability to capture changes made outside the Django admin interface. It provides tips for selective tracking, proper usage of EventModelAdmin, and taking initial snapshots of existing data for accurate change logging.

This video discusses the upcoming features, performance improvements, and other changes in Python 3.14, including the tail call interpreter, JIT compiler, and free threading. It also covers minor updates and deprecations, providing a comprehensive overview of the new release.

This article discusses an approach to render a Shakespeare quote quickly by optimizing a mathematical formula using RPython and C. Key optimizations include using quadtrees, a simple optimizer with interval analysis, and a "demanded information" optimization that removes operations by only keeping the sign of the result.

This post provides a step-by-step guide to creating a desktop sticky notes application using PySide6, covering GUI setup, styling, and adding functionality like movable notes and a system tray icon. It explains how to build a simple, frameless note window, enable dragging, and manage multiple notes, with code examples for each stage.

This article reproduces the word2vec model using JAX, explaining the CBOW architecture and its implementation. It trains word embeddings and demonstrates how to find word similarities and analogies using the trained model, comparing it to modern text embeddings used in LLMs.

This post explains how to extract GPS coordinates from a photo using Python and plot them on a map, using libraries like Pillow, ExifRead, and Folium. It challenges the reader to analyze the location of a USAID nutrition pack to determine if the aid is being distributed appropriately.


Python Jobs of the Week

Build next-gen AI tools transforming private equity diligence—now hiring a full-time Software Engineer in NYC. Work across the stack (FastAPI, Next.js, PostgreSQL, PyTorch), implement cutting-edge techniques like RAG and Text-to-SQL, and help scale systems used by top investors. 2+ years' experience required.


Interesting Projects, Tools, and Libraries

Memo is a simple command-line interface (CLI) tool for managing your Apple Notes (and eventually Apple Reminders). It’s written in Python and aims to offer a fast, keyboard-driven way to create, search, and organize notes and reminders straight from your terminal.

Transformers-compatible library for applying various compression algorithms to LLMs for optimized deployment with vLLM.

The NCA Toolkit API eliminates monthly subscription fees by consolidating common API functionalities into a single FREE API. Designed for businesses, creators, and developers, it streamlines advanced media processing, including video editing and captioning, image transformations, cloud storage, and Python code execution.

Towards a Unified Platform, Dataset and Benchmark for Scalable and Generalizable Robot Learning.

A special build of curl that can impersonate Chrome & Firefox.

Scaling any Modality Model Training to any Accelerators with PyTorch native Training Framework.

Eliciting Factual Medical Reasoning Steps in LLMs via Knowledge Graphs.

Single File, Single GPU, From Scratch, Efficient, Full Parameter Tuning library for "RL for LLMs"

A single interface to different Agent frameworks.

An open-source, code-first Python toolkit by Google for building, evaluating, and deploying sophisticated AI agents with flexibility and control.

Open Source Algo Trading Platform for Everyone.

Evaluate and improve your Retrieval-Augmented Generation (RAG) pipelines with open-rag-eval, an open-source Python evaluation toolkit.


Upcoming Events and Webinars

Microsoft is holding an AI Agents Hackathon, and they want to see what you can build with Python! There will be 20+ live streams showing you how to build AI agents with Python using popular agent frameworks and Microsoft technologies. Then, you can submit your project for a chance to win prizes, including a Best in Python prize!

The April edition of IndyPy will showcase leading-edge tools designed to supercharge your development workflow. From uv, a blazing-fast Python package manager, to a live comparison of the top AI coding assistants, you'll see firsthand how to eliminate bottlenecks, simplify setup, and code smarter.

There will be following talks

  • Large Scale Classification with LLMs (1000+ categories)

  • Going from Notebooks to Scalable Systems

There will be a talk, Creating real-time navigable 3D animations with Python.

There will be following talks

  • Why I hate JAX, but still choose to use it.

  • Hybrid Intelligence in Medical Image Segmentation

There will be following talks

  • GenAI for Code Migration: My Journey From Supermassive Black Holes to Supermassive Codebases

  • Hold on, where's my context...?

There will be following talks

  • Practical Python Parsing with the ast Standard Module

  • EU AI Act – The Abridged Version


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.