- Python Weekly
- Posts
- Python Weekly (Issue 714 September 18 2025)
Python Weekly (Issue 714 September 18 2025)
Rahul Chaudhary
September 18, 2025
Welcome to issue 714 of Python Weekly. Let's get straight to the links this week.
Don’t be the one putting your company at risk 😳
Most meetings include confidential data - from project details to client information.
But not every AI notetaker protects that information with the security it deserves.
If the AI notetaker you’re using trains their AI models on your team’s conversations, you could be putting your company at risk without realizing it.
This AI Meeting Notetaker Security Checklist helps you avoid that.
In just two minutes, you’ll learn the 7 checks to ensure your team’s AI meeting notes stay private and secure.
Don’t let your meetings become someone else’s dataset.
Articles, Tutorials and Talks
The video provides an in-depth, hands-on tutorial for building a resilient, AI-assisted Reddit scraping pipeline in Python, covering everything from Jupyter prototyping and LangChain agents to a Django-based background worker architecture. It teaches viewers to automate web scraping, integrate Google’s Gemini LLM for query refinement, and store structured results in PostgreSQL, suitable for scalable, real-world data workflows.
A hitchhiker's guide into LLM post-training.
A course of learning LLM inference serving on Apple Silicon for systems engineers: build a tiny vLLM + Qwen.
The video demonstrates how to keep Python code for AI agents clean and maintainable by applying design patterns like Chain of Responsibility (for modular pipelines), Observer (for agent logging and context), and Strategy (for pluggable agent personalities). These patterns help decompose logic, improve scalability, and ensure testability for complex AI workflows.
Python-BPF is a new open-source library that lets developers write eBPF programs directly in Python, compiling them into kernel-ready object files instead of embedding C code. By using Python decorators and standard Python tools, this approach simplifies development and expands eBPF accessibility, though the project is still experimental and not yet production ready.
The post describes an experiment in animating a mosaic of vintage 90s GIFs collected from the GeoCities archive, using HTML Canvas for random, lively playback. It celebrates the playful aesthetics of early web graphics and highlights the technical and nostalgic joy of reintroducing these classic GIFs into a modern browser setting.
This video shows how to tune XGBoost models with Optuna while maximizing speed using XGBoost 3.0’s GPU acceleration for 5–15x faster training. He explains why cross-validation is crucial, recommends smart tuning practices, and demonstrates how Optuna’s visualizations help identify impactful hyperparameters in real-world tabular data workflows.
An LLM that can converse in English & item IDs, and make recommendations w/o retrieval or tools.
PyCon Australia 2025 talks videos are available now.
LLM inference is often nondeterministic even with temperature set to zero, primarily due to batch-size-dependent kernel behaviors that change results based on server load rather than randomness or floating-point issues. The solution is to use batch-invariant kernels, ensuring reproducible outputs even in high-concurrency environments, which is now possible but may come with some efficiency trade-offs.
The guide demonstrates how to use the OpenAI Agents SDK’s Session object to manage short-term memory in AI agents, enabling context trimming and compression for efficient, coherent, and cost-effective multi-turn conversations. Effective session memory ensures agents maintain relevant history across turns while reducing noise, latency, and error risk in longer interactions.
The post details major upgrades that allow models like OpenAI’s GPT-OSS to run, fine-tune, and scale efficiently, including zero-build kernels, 4-bit MXFP4 quantization, tensor and expert parallelism, dynamic layerwise caching, and continuous batching. These improvements cut memory usage, boost speed, and enable larger models to run on affordable hardware, making cutting-edge techniques easily accessible through the transformers library.
Interesting Projects, Tools, and Libraries
Production-ready GraphRAG with multi-modal indexing, AI agents, MCP support, and scalable K8s deployment
Like requests for mathematical computing, making complex math feel simple.
The Modern Async Security Framework for FastAPI.
detroit is a Python implementation of d3js.
A lightweight, from-scratch, object-oriented Python package implementing classic numerical methods.
Enterprise-Grade Graph RAG for Secure, On-Premise AI with Verifiable Attribution.
A Textual User Interface for interacting with Atlassian Jira from your shell.
A meta-agent framework to build high-performance multi-agent systems.
Scaling Up Reasoning Patterns and Interaction Turns for Visual Search.
A toolkit for building PostgreSQL wire-compatible databases in Python, powered by Rust for performance and concurrency.
Run Python like a local function from Go — no CGO, no microservices.
Build data processing and data analysis pipelines that leverage the power of LLMs.
New Releases
This release includes the following announcements:
Experimental AI-powered hover summaries with Pylance
Run Code Python Snippet AI tool
Python Environments extension improvements, including pipenv support
Upcoming Events and Webinars
There will be a talk, From Queries to Models: Python + Exasol Analytics Database in Action.
There will be following talks
Reinforcement learning for games at Supercell
Estimating weight of cars from their dimensions
Marimo - a reactive, git-friendly Python notebook
There will be a talk, MVP – Minimal Viable Prompting Game Day.
There will be following talks
Prototype: A framework that lets analysts focus on ideas, not code
Async to Distributed: Concurrency Patterns for Data Processing
There will be following talks
How to Train Your LLMs?
Data and Training in AI Models - Why Empathy Matters
There will be following talks
Building Multimodal Generative AI Applications with Mosaic AI
Elevating Prompting Engineering Skills: Prompt Patterns and Meta-Prompting
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.