| • Data science and its role in industry | |
| • Challenges with data cleaning and preparation | |
| • Limitations of Large Language Models (LLMs) in data science | |
| • The concept of "data engineering" vs. "data science" | |
| • Philipp Burckhardt's background and work as a data scientist | |
| • The standard library project, stdlib.io, and its mission to make the web a preferred environment for numerical computation | |
| • The speaker's background in economics and statistics led them to work with JavaScript for data modeling and scraping | |
| • The growth of the JavaScript ecosystem and its impact on the speaker's work | |
| • The development of a project involving sentiment analysis and topic modeling, which relied on MongoDB and JavaScript | |
| • Meeting Athan Reines through GitHub, who hired the speaker for a summer job and led to a long-term collaboration | |
| • The creation of a third-party standard library for JavaScript, initially focused on math and numerical computing | |
| • Standard library limitations in JavaScript historically | |
| • Creation of stdlib.js to provide a rigorous set of tools and functions | |
| • Focus shifted from utility functions to numerical and statistical computing infrastructure | |
| • Proliferation of JavaScript runtimes (Bun, Deno) and impact on standardization | |
| • Design goals: fully decomposable, modular, and coherent user experience | |
| • Library size: 3,800 repositories with 150+ special math functions and 200+ general utilities | |
| • Consumption options: install individual packages or the whole library | |
| • TypeScript declarations for all packages | |
| • ESM support with tree shaking | |
| • Decomposability of deployment and modularity | |
| • Use of var instead of const and let declarations | |
| • Potential transformation scripts to update code | |
| • Bike-shedding debate on const vs let usage | |
| • Current project status: ongoing development, contributor base, performance improvements | |
| • Architecture for decomposability: | |
| • Node.js module resolution algorithm | |
| • Explicit dependencies | |
| • Tooling pipe chain with GitHub Actions workflows | |
| • Monorepo organization and automation | |
| • Current state of JavaScript modules and packages | |
| • Transition from CommonJS to ESM (ECMAScript Modules) and its complications | |
| • Philipp's stance on waiting out the transition rather than rushing into new features | |
| • Importance of backward compatibility and maintainability in a standard library project | |
| • Size and scope of the stdlib project, including 50,000+ commits and 360 open issues | |
| • Prioritization and maintenance process for the project | |
| • Need for community involvement and potential corporate backing to take the project to the next level | |
| • Future plans to explore using stdlib as a foundation for numerical computing in JavaScript | |
| • Collaboration with other projects and libraries to standardize the ecosystem | |
| • stdlib project provides a collection of libraries for numerical computing in JavaScript | |
| • Libraries include tools for data manipulation, statistical tests, and client-side computations | |
| • Useful for full-stack developers working with React, Express, and MongoDB | |
| • Project has office hours for contributors and users to ask questions | |
| • Upcoming plans include finishing linear algebra functionality and automating the process of creating multi-dimensional arrays | |
| • Google Summer of Code will be a major contributor to the project this year |