SELECT * SQL; JavaScript 30; The Year Of Embeddings

While you’re off self-hosting the resources from the last Drop, perhaps take a moment to assess what you may want to dig into this year. Do you want to learn/become more proficient in SQL? JavaScript? Go? R? One or more frameworks?

If so, perhaps one of today’s resources will catch your eye! Let’s take a look at three areas you may want to level up a bit in this new year filled with new opportunities!

TL;DR

This is an AI-generated summary of today’s Drop.

Yay! It started adding links on its own again!

Here is a concise summary of the post:

1. SELECT * SQL: The blog post highlights the importance of SQL in data science and app development. It introduces Select Star SQL, an online platform created by Zi Chong Kao, a former data scientist at Quora. The platform offers practical, hands-on learning through a variety of exercises, using a unique dataset about Texas death row inmates. The platform uses client-side databases, primarily running a SQLite database directly in the user’s browser.
   
2. JavaScript 30: The post recommends learning JavaScript and introduces JavaScript 30, a resource created by Wes Bos that challenges users to build 30 things with vanilla JavaScript in 30 days. The course is designed for developers of all levels and emphasizes hands-on learning through coding exercises and projects. It’s free and encourages users to commit to the course publicly and invite friends to join.
   
3. The Year Of Embeddings: The post predicts that 2024 will be the year of embeddings in tech. It provides a list of resources for learning about embeddings, including YouTube tutorials, research articles, and community discussions. The resources cover a range of topics, from the basics of embeddings to more advanced topics, and are suitable for individuals at any level of responsibility within a technical organization.

---

SELECT * SQL

As we’ve noted in many-a-Drop, SQL is almost a requirement for doing anything in app- or data science-land, these days. Even if there are abstraction layers (such as R’s {tidyverse}), it’s very useful to know how the SQL sausage is being made, as you may need to — at some point — optimize it on your own.

Select Star SQL is as an online dojo for those seeking to master SQL. It is a creation born from the challenges and inspirations of Zi Chong Kao, a former data scientist at Quora. During Kao’s tenure at Quora, he saw there was a gap in the availability of effective SQL learning resources that were not only free and accessible, but also demonstrated a genuine commitment to learners. This unmet need spurred him to build the site.

The site has a strong emphasis on practical, hands-on learning. It is structured to provide folks with a cornucopia of exercises that are not only realistic but also substantial. These exercises are meticulously crafted to guide learners through the intricacies of SQL, introducing them to progressively sophisticated techniques. The journey through SQL is not just theoretical; it is anchored in the exploration of an intriguing dataset, making the learning experience both engaging and informative.

The choice of dataset for Select Star SQL is unique and thought-provoking (I’m very partial to “data journalism”-ish datasets). It revolves around a comprehensive compilation of data concerning Texas death row inmates, spanning from 1976 to the present. This dataset, extracted from the Texas Department of Criminal Justice website and painstakingly cleaned, stands as one of the most complete repositories of Texas death row data available on the internet. This dataset not only serves as a tool for learning SQL but also provides a window into a significant socio-legal aspect of Texas’s history.

From a technical standpoint, Select Star SQL adopts an innovative approach to circumvent the costs associated with web hosting while still providing an interactive database experience. The platform utilizes client-side databases, primarily running a SQLite database directly in the user’s browser thanks to SQLite’s epic WASM powers. Not having to rely on a perfect workstation setup goes a long way into making Select Star SQL far more accessible than many other (legacy) SQL tutorials.

Please poke around the details of the platform to catch the kudos Kao heaps upon those who have helped make the resource as great as it is.

JavaScript 30

Longtime readers know one of my “broken records” is that you should learn and become proficient in JavaScript. I still hold by that in 2024, so let’s take a look at one way you can level up those skills.

JavaScript 30, created by Wes Bos, is a unique resource that challenges folks to build ‘30 things with vanilla JavaScript in 30 days’. It’s a fantastic way to learn JavaScript, regardless of your proficiency level.

The site is all about hands-on learning. It’s not just about reading or watching tutorials; it’s about actively engaging with coding exercises and projects. This approach helps you apply theoretical knowledge to practical scenarios, promoting a deeper understanding of concepts. As the author puts it, “How do you get better? Build things. Lots of things. Build 1,000 things. Keep it up and don’t stop. Seriously”. This philosophy is at the heart of what Wes built, making it a great resource for those of us who learn best by doing.

The resource is also designed for developers of all levels. Whether you’re a beginner looking to get comfortable with JavaScript fundamentals or an intermediate developer wanting to work in the DOM without a library, this course has got you covered. It’s also a great way to get familiar with ES6 concepts, which are widely used in modern JavaScript development.

JavaScript 30 is also free. This makes it a low-risk investment for anyone interested in strengthening their JavaScript skills. Despite being free, the course is well-produced and offers a wealth of knowledge. Each lesson guides you through building pure vanilla JavaScript projects, no frameworks, no libraries, just JavaScript.

And, you won’t be learning alone. When you sign up, you’re encouraged to commit to the course publicly and invite some friends to join you. This creates a sense of accountability and camaraderie that can make the learning process more enjoyable and motivating.

It’s a fun, engaging way to really wrap your noggin’ around JavaScript. You will not regret the time spent.

The Year Of Embeddings

While “AI” ate the tech zeitgeist in 2023, 2024 is most certainly the year you will not be able to get away from embeddings. Now, I’m not saying you need to become an “AI” engineer/data scientist. Folks at any level of responsibility within a technical organization should be familiar with this concept.

For those new to the concept of embeddings and looking to learn about them in 2024, there are several resources that can help you get started and understand how to use them effectively.

I went through a whole bunch of “embeddings”-related items I’ve shoved into Raindrop over 2023 and tucked them into the list below. Please note that, wherever possible, I try to showcase “new faces” and known talented individuals vs. rely on major brands or sites. However, some of those “majors” are difficult to avoid.

YouTube Tutorials

Fair warning: I consume video/audio content at 2x speed, so apologies if any of the video-ish ones may not work as well at plain speed.

• NLP Roadmap 2024: Step-by-Step Guide offers a visual and step-by-step overview that can be particularly helpful for beginners. This resource covers the basics and provides a roadmap for further learning in natural language processing (NLP), which is a common area where embeddings are used.
• A Beginner’s Guide to Vector Embeddings provides a high-level primer on vectors, vector embeddings, and vector databases, which could be a good starting point for those new to the topic.
• Creating Embeddings with OpenAI: A Beginner’s Guide might offer insights into how to create embeddings specifically using OpenAI’s tools, which could be particularly relevant given the rise of large language models and their applications. OPENAI TAX REQUIRED.
• A Complete Overview of Word Embeddings gives a comprehensive look at word embeddings, which are crucial for natural language processing tasks.
• Embeddings: What they are and why they matter discusses the importance of embeddings and their role in machine learning, which could help in understanding their broader significance.
• AI embeddings explained: understanding the basics and beyond aims to explain the basics of AI embeddings and delve into more advanced topics, suitable for those who want to deepen their understanding. (This one is especially good for consumers or technical managers).
• Understanding embeddings in one minute #GenerativeAI offers a quick overview, which might be useful for those looking for a brief introduction or refresher. (Don’t play this at 2x; only tap it if you’re 100% new to embeddings.)
• Tutorial 1: Comparing Word Embedding Models provides a comparison of different word embedding models, which can be beneficial for understanding the nuances and differences between various approaches. (This is an older-ish resource, and part of larger curriculum. But, I think it helps show some of the ways we got where we are in this space.)

General Research and Articles

• Improving Text Embeddings with Large Language Models In this paper, researchers from Microsoft introduce a novel and simple method for obtaining high-quality text embeddings using only synthetic data and less than 1k training steps. (Very academic-y.)
• The Full Guide to Embeddings in Machine Learning on Encord’s blog is a comprehensive resource that explains the importance of high-quality training data and how AI embeddings can improve it. (This guide also discusses best practices for using AI embeddings, which can be valuable for beginners.)
• Getting Started With Embeddings on Hugging Face’s blog provides a tutorial with a Notebook Companion that can be opened in Google Colab. (This resource is great for understanding the basics of embeddings and includes practical examples.)

Blogs and Community Discussions

• Detailed guide on training embeddings on a person’s likeness on (of all places) Reddit is a decent guide on how to train embeddings with textual inversion on a person’s likeness. This can be particularly interesting for those looking to apply embeddings in image processing or generative art.
• An intuitive introduction to text embeddings on Stack Overflow’s blog offers an accessible introduction to text embeddings, which are key to large language models (LLMs) and convert text into vector coordinates.

FIN

If you have any fav ones, drop a note in the comments or reply to this post on Mastodon. Remember, you can follow and interact with the full text of The Daily Drop’s free posts on Mastodon via @dailydrop.hrbrmstr.dev@dailydrop.hrbrmstr.dev ☮️