AI applications that produce human-like text, such as chatbots, virtual assistants, language translation, text generation, and more, are built on top of Large Language Models (LLMs). If you are deploying LLMs in production-grade applications, you might have faced some of the performance challenges with running these models. You might have also considered optimizing your deployment with an LLM inference engine or server.

Hey there! We're back for our third edition of Tips and Tricks, our new mini series where we share some helpful insights and cool tech that we've stumbled upon while working on technical stuff. Catch up on the previous posts: All of our posts are super short reads, just a couple of minutes tops. If you don’t like one of the posts, no problem! Just skip it and check out the next one. If you enjoy any of the topics, I encourage you to check out the "further reading" links.

There are several ways to handle load spikes on a service. However, these methods are not cost-effective: you either pay for resources you don't use, or you risk not having enough resources to handle the load. Fortunately, there is a third way: horizontal autoscaling. Horizontal autoscaling is the process of dynamically adjusting the number of instances of a service based on the current load. This way, you only pay for the resources you use, and you can handle load spikes without any manual intervention.

Hey there! We're back for our third edition of Tips and Tricks. As we said in our first posts on Drizzle ORM and Template Databases in PostgreSQL, our new Tips and Tricks mini blog series is going to share some helpful insights and cool tech that we've stumbled upon while working on technical stuff. Today's topic is short and sweet. It'll be on CPU utilization and what that metric indicates. If you enjoy it and want to learn more, I encourage you to check out the "further reading" links.

Today, we’re excited to share that Serverless GPUs are available for all your AI inference needs directly through the Koyeb platform! We're starting with GPU Instances designed to support AI inference workloads including both heavy generative AI models and lighter computer vision models. These GPUs provide up to 48GB of vRAM, 733 TFLOPS and 900GB/s of memory bandwidth to support large models including LLMs and text-to-image models.

Hey there! We're back for our second edition of Tips and Tricks. As we said in our first post on Drizzle ORM, our new Tips and Tricks mini blog series is going to share some helpful insights and cool tech that we've stumbled upon while working on technical stuff. Today, we're going to talk about the template databases of PostgreSQL. Remember, these posts will be super short reads. If you don’t like the topic of one of the posts, no problem! Just skip it and check out the next one.

To keep up with everything happening in the world of artificial intelligence, it helps to understand and grasp key terms and concepts behind the technology. In this introduction, we are going to dive into what is generative AI, looking at the technology and models they are built on. We'll discuss how these models are built, trained, and deployed into the world.

Hey there! At Koyeb, we really like diving into technical stuff. But here’s the thing: not every cool thing we stumble upon or think about needs a massive blog post. And honestly, not everything we’re into is directly related to what Koyeb does or about infrastructure in general. So, we’ve got an idea: what if we start sharing these bits and pieces with you in a series of really short blog posts?

toddl.co is an all-in-one booking platform for kids' activities in Spain. Offering more than 2,000 classes, camps, and events to over 17,000 monthly visitors, toddl.co is on a mission to help parents navigate the complex world of extra-curricular activities and to help activity organizers manage and grow their businesses. For businesses, toddl.co streamlines bookings, payments, and client management. This year, toddl.co plans to serve over 15,000 businesses and 50,000 families.

Retrieval-augmented generation (RAG) is an AI framework and powerful approach in NLP (Natural Language Processing) where generative AI models are enhanced with external knowledge sources and retrieval-based mechanisms. These appended pieces of outside knowledge provide the model with accurate, up-to-date information that supplements the LLM’s existing internal representation of information. As the name suggests, RAG models have a retrieval component and a generation component.