[Dev Catch Up #54] - Mistral OCR, Deepgram's Nova-3 Medical, FoxConn's FoxBrain, TypeScript 5.8, GoatDB, Lynx, Yaak, and much more!
Bringing devs up to speed on the latest dev news
Welcome to the 54th edition of DevShorts, Dev Catch Up!
For those who joined recently or are reading Dev Catch Up for the first time, I write about developer stories and open source, partly based on 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
How to do Local AI Code Generation and Completion using Open-source Llama.cpp?
Join 6500+ developers to hear stories from Open source and technology.
Must Read
OCR models seem to be emerging all the time and now Mistral AI released Mistral OCR, an OCR API that extracts text, images, tables and math formulas from PDFs and images. It works with many languages, processes up to 2000 pages per minute, and gives you JSON output. If you work with sensitive data, you can self-host it. Read more about Mistral OCR here.
Foxconn has unveiled its first large language model named FoxBrain. It is built on Meta’s Llama 3.1 and trained in four weeks using 120 Nvidia H100 GPUs. It is designed to handle many tasks like analyzing data, and documents, doing math, and coding etc. To Learn more, read the detailed blog post on FoxBrain.
Managing separate codebases for web and mobile is a real pain for developers. Lynx solves this problem by letting you create native UI for both platforms from a single codebase. ByteDance developed it, and it powers apps like TikTok. Lynx builds fast, responsive apps that look perfect and launch instantly. Check this blog to know more about Lynx.
Models are expanding across domains, and now Deepgram has released its next healthcare model Nova-3 Medical. It offers low latency, HIPAA compliance, and security while ensuring high transcription accuracy. It understands medical terms and ideal for clinical documentation, EHR integration, and telemedicine transcription. Check this article for more information on Nova-3.
OSS Highlight of the Week
GoatDB is a Light Weight No DB solution designed for Deno, React, and low-friction deployments. It runs entirely client-side, no need of infrastructure. It supports offline access, fast performance, and real-time syncing. React devs will like GoatDB's hooks that handle state management, syncing, and saving data automatically. It is secure, scales easily, and makes app development simpler.
Good to know
While API clients like Postman dominate, new tools are emerging. As a developer constantly switching between platforms, I found Yaak interesting, it is a desktop API client, that handles REST, GraphQL, SSE, WebSocket, and gRPC with features like Postman import, OAuth. Worth checking out if you're tired of juggling multiple clients.
Text-to-speech models are everywhere, but Spark-TTS caught my attention. With zero-shot voice cloning in Chinese and English, it offers flexible speech synthesis. Powered by Qwen2.5, it supports multiple languages with easy voice customization. Check this out if you need flexible speech synthesis.
Being a Python developer, I know how difficult it is to deal with module dependencies. Tach offers a CLI tool that visualizes module dependencies, verifies imports, and runs only affected tests. If you are working with larger projects, Tach may help you maintain your code clean and modular.
We all have definitely used S3 buckets in our projects, and securing these buckets is more important than ever. This blog post explains how to restrict Amazon S3 bucket access to a specific IAM role using policy configurations, ensuring only authorized users can access sensitive data.
Many JavaScript developers struggle with implementing design patterns effectively. This GitHub repository provides simple implementations of common patterns like Singleton, Factory, and Observer. A go-to resource for understanding software design concepts.
Every one of us definitely worked with Linux OS. But many of us do not know the inner working of Linux OS. I found this blog post explores the fundamentals of Linux kernel development, explains about process scheduling, memory management, system calls, and hardware interaction and much more.
As a developer, I know how challenging it can be to create surveys that work across different web platforms. Check SurveyJS library that lets developers render JSON-based forms, collect responses, and store submission data in any database. Read more about SurveyJS in this page.
Notable FYIs
TypeScript keeps evolving with every major release, and version 5.8 is no exception. It brings support for
require()
in ES modules, a Node 18 resolution mode, granular type checks, and performance improvements. If you're building modern JS apps, the new compiler flags can help streamline your workflow. Definitely worth checking this post if you're looking to stay current with TypeScript.Integrating JavaScript and Rust can definitely be tricky due to performance overhead. This blog post reveals a new way to eliminate serialization overhead by rethinking data transfer strategies. If you are working with both languages, it’s a must-read for efficiency gains.
Working with large-scale data definitely requires specialized tools for visualization. Perspective is an open-source analytics component. Definitely check the tool if you need interactive dashboards for massive datasets.
As a developer, I’m pretty excited about Next.js’s experimental Node.js runtime for Middleware. You can now tap into Node.js modules directly in your middleware for more complex logic, just by flipping a config flag. It’s still canary-only, but if you want full Node.js capabilities in Next.js, this is worth a try. Read more about this feature in this page.
We all know how crucial logging is, without it, we’re flying blind. If you’re a JavaScript/TypeScript dev, give LogTape a try. It has zero-dependency, runs on Node.js, Deno, Bun, edge, and browsers, and supports structured logging, customizable verbosity, dynamic messages, and easy extensibility.