[Dev Catch Up #49] - RLHF Short Book, o3-mini high, Temporal Object in JavaScript, Right Docker image for Node.js, DexieJS, Deep Research agent, and much more.
Bringing devs up to speed on the latest dev news from the trends including, a bunch of exciting developments and articles.
Welcome to the 49th edition of DevShorts, Dev Catch Up!
For those who joined newly or are reading, Dev catch up for the first time. I write about developer stories and open source, partly from my work and experience interacting with people all over the globe.
Thanks for reading Dev Shorts! Subscribe for free to receive new posts and support my work.
Some recent issues from Dev Catch up:
How to use Postgres as a Vector Database with BGE Embedding model
Apollo video LLM, Gemini 2.0, lla file explorer, Database migration
Join 6000+ developers to hear stories from Open source and technology.
Must Read
Reinforcement Learning has been the topic of discussion throughout the last few weeks. Recently, I discovered this book, which explains the techniques and details of popular algorithms. It fills in the important gap between the theory and practice of RLHF, nicely binding everything together.
Previously in our issues, I have written a lot about LLM-as-a-Judge technique that came as a breakthrough in the development of reasoning capabilities of a model. However while snooping around papers on AI developments, I found something interesting on the same. Have a look at this paper that proposes EvalPlanner, a preference optimization algorithm for Thinking-LLM-as-a-Judge.
People are going crazy over the launch of o3-mini, the fastest, smartest, and most cost-effective reasoning model yet from OpenAI. With its intelligence, capability, and speed, it can solve complex coding problems, simulated 3D figures, answer complex questions, and much more. Have a look at this x.com thread depicting the same and learn more about the model from the official launch article.
OSS Highlight of the Week
Catgrad, is a deep learning framework that uses category theory to statically compile models into their forward and backwards passes. Learn more about it from its official web page and check the project out from the GitHub page.
Hope you are enjoying this edition of our newsletter so far! Support us by following our LinkedIn and X pages.
Your support is highly appreciated!
Good to know
While Mistral has made quite the name in the early days of LLM development, the launch of Mistral Small 3 made it back to the spotlight. It is a latency-optimized 24B-parameter model and competitive with larger models such as Llama 3.3 70B or Qwen 32B. Learn more about this Apache 2.0 open-source model from the official launch article.
Not all developments are happening with AI, a lot of it is happening in the field of JavaScript as well. As a frontend developer, you might have faced complexity with dates and times but it will be completely simplified and modernized with the introduction of the Temporal object. This article will shed more light.
Containerizing an application has always been challenging, especially with picking the right base image. Docker is the defacto choice for containerization, and the evaluation of the best image among the many depends on several factors. This article discusses picking the right image for a node.js application.
If you work with Node.js on a daily basis, then you can encounter a problem while using
fetch
for amultipart/form-data
request. Here is an article that focuses on the solution to this problem.As a dev, making a request to Open RSS feeds is not uncommon. But do you use the Open RSS service in a proper guided way so that the service won’t get compromised for other users. Here is a detailed guide for a developer wanting to use Open RSS feeds.
Notable FYIs
While there are multiple available agentic capabilities, the one that excited me the most is deep research. Google introduced it in their gemini chatbot a few days earlier and now OpenAI came up with it in ChatGPT, where the agent uses reasoning to synthesize large amounts of online information and complete multi-step research tasks for you. Learn more about it from here.
useState
hook has always been buggy in React and combined with manual loading states and complex setups for a simple async query made life difficult for a developer. DexieJS solved this state management problem with live queries anduseActionState
hook. This guide shows it with a step-by-step implementation.I bet you have heard about or used Golang once in your life. It is a language that can handle everything from application development to infrastructure management. Here is an article that shows why choosing Golang as a programming language for getting started is not a bad choice.
What other text editors do you know other than VSCode? Many will answer Sublime Text. It is a great editor and this article explains why it is still relevant in 2025.
Most of you use Google Chrome for daily web browsing. But do you know that you can chat with AI assistants from Chrome’s address bar? It can be done with a simple hassle-free trick and this Labnol article helps you do it.
That’s it from us with this edition. We hope you are going away with a ton of new information. Lastly, share this newsletter with your colleagues and pals if you find it valuable and a subscription to the newsletter will be awesome if you are reading for the first time.