About:

Vincent is a senior data professional passionate about algorithmic systems and open-source, known for his calmcode.io project and involvement in data conferences.

Website:

Specializations:

Interests:

Algorithmic systems Open-source packages Python notebooks Data projects

Incoming Links:

Outgoing Links:

Simon Willison
Subscribe to RSS:
The blog post discusses the Titanic dataset, commonly used in data science to predict survival based on passenger properties. It highlights a unique approach by focusing on the 'name' feature, which can yield comparable model accu...
A comparative analysis of various programming languages and libraries for simulating a prisoner problem, highlighting performance metrics and the importance of understanding code design.
Optimizing GridSearchCV for specific use-cases can significantly enhance performance in Ridge and Logistic Regression models, but careful validation of AI-generated code is essential.
The Sock Drawer Paradox presents a mathematical puzzle involving the probability of drawing red socks from a drawer containing red and blue socks. Given that the probability of drawing two red socks is 1/2, the puzzle explores the...
The Borsuk-Ulam theorem illustrates that antipodal points on a sphere have the same value, demonstrated interactively through Python visualizations.
The post discusses the limitations and frustrations of using Claude, an AI coding assistant, particularly in customizing workflows and permissions. The author shares an experiment involving a bespoke agent that allows for more con...
The author discusses the evolution of chatbot tools, specifically focusing on Rasa's transition from traditional state machine forms to a more flexible approach using LLMs (Large Language Models). The text highlights how the new s...
The term 'sharding' in databases originates from the lore of Ultima Online, where a gem's shattering led to the creation of multiple game servers.
Using extreme conciseness in prompts for Huber regression explanations can yield more detailed results compared to grammatically correct prompts.
Benchmarking the diskcache library reveals how a custom JSONDisk class with zlib compression can optimize data storage, especially for text-heavy applications.
The blog post discusses how to maintain code portability across different database backends when using SQLAlchemy. It highlights the importance of avoiding dialect-specific code that could break compatibility, particularly when bu...
The post discusses how to ensure SQLAlchemy code remains compatible across different database backends by avoiding the use of backend-specific features. It highlights the JSON support in PostgreSQL and the lack of it in SQLite, an...
The post discusses the use of high-resolution moving SVGs, which are similar to GIFs but utilize SVG technology. It explains how to implement these SVGs in GitHub README.md files using tools like asciinema and svg-term-cli. The au...
The blog post discusses Daytona, a cloud provider designed for sandbox use-cases, particularly for running code generated by LLMs in a secure environment. It highlights the straightforward setup process, the ability to create cust...
Sediment charts visualize the evolution of GitHub repositories, revealing code changes and project health over time, particularly in Python projects.
The post discusses the concept of 'deliberate practice' as proposed by psychologists, emphasizing that effective practice involves setting goals, focusing, and reflecting on weaknesses. The author introduces the idea of 'deliberat...
The author shares their experience using the AI/LLM tool Cline through its VSCode extension. They highlight the unique 'plan' and 'act' modes that enhance context awareness during coding tasks. The author appreciates the transpare...
The post discusses the features and benefits of using 'just', a command runner that simplifies task automation similar to 'make'. It highlights functionalities such as creating a global just file for reusable commands, managing wo...
Different methods to simplify the expression $$ \frac{\sin^2 x}{1 - \cos x} - 1 $$ are explored, ultimately leading to the solution $$ \cos x $$.
Wget is a powerful tool for downloading entire websites, with specific command-line options that enhance its usability for offline access to documentation.
The blog post introduces TaskiPy, a tool that allows users to define task automation directly in their pyproject.toml file, serving as an alternative to Makefiles. It explains the basic setup, including how to add tasks, chain com...
The author discusses the evolution of their approach to open-source projects, distinguishing between personal tools meant for individual use and those intended for broader distribution. They emphasize the importance of using platf...
The post discusses the challenges of context in image generation, using the example of a 'Starcraft' marine riding a vulture. It compares two generated images based on a prompt and highlights the difficulty in determining which im...
The blog post discusses a YouTube video showcasing the customization of a Framework laptop with a custom mechanical keyboard. It highlights the repairability and hackability of the Framework platform, emphasizing features like a g...