[Dev Catch up #12] - Google Gemini launch, AWS re:invent, Microsoft Loop, and more

Bringing devs up to speed on the latest dev news from the trends including the launch of Google Gemini, announcements from AWS re:invent, and a bunch of exciting developments and articles.

The evolution of tech with new developments and happenings is constant. And as usual, DevShorts is back with another issue to simplify your digests from the community. Like our last ones, this issue also covers the unique stories trending in our developer circle, along with a look at new open-source projects, tutorials, conference news, and much more.

Must Read

• Google took the internet by storm after jumping in on the competition of building generative AI models. While giving direct competition to OpenAI with the launch of a conversational AI assistant named Bard, they recently introduced the Gemini model which is presumed to be a far superior model to the newly launched GPT4 model introduced by Open AI. The largest version of the model exceeds GPT4 on 30 of the 32 academic benchmarks that are commonly used for evaluating Large Language Models. Currently, two more versions of this model are out. One is specially tuned in English to work with the AI assistant chatbot Bard and is available in most countries. The smallest version of the model is designed so that it can be locally run on any consumer device. Learn more about this model from this detailed article from Arstechnica, which explains how Gemini differs from the already existing state-of-the-art GPT4 and how it defines a new era in the world of computing.  

• Amazon just ended its biggest cloud conference of the year, AWS re:Invent with a lot of spectacular announcements. With all the announcements, Amazon reaffirmed its support for the advancement of tech with the help of AI and ML. They are deploying AI tools and services to hold the largest share of the cloud market. Throughout re:invent, there are a bunch of announcements of new offerings. Three serverless offerings to manage Aurora, Elasticache, and Redshift serverless services and the upgradation of S3 object storage service to Amazon S3 Express One Zone. Offerings on AI made headlines with the launch of new AWS Trainium chips for AI models. For the customers, there is the launch of Amazon Q, which is an AI-powered chatbot and guardrails for Amazon Bedrock that will help companies define and limit model use. From Neptune Analytics to Clean Rooms ML, re:invent was filled with launches regarding AI, and learn all about these exciting launches and offerings from this article written by TechCrunch.

• Microsoft joins the battle of collaborative workspace applications with the launch of Loop. Similar to Notion’s interface, Loop lets you use and create flexible and collaborative workspaces and pages. Unlike Notion, it has an interactive UI with the support of Microsoft applications like Teams, Outlook, etc., which will eliminate the need to switch between apps. Loop is also packed with Microsoft’s AI-powered Copilot assistant which can help with drafting texts and summarizing pages in the app. Learn more about Loop in detail from this article posted by The Verge and from its official product page. 

• Generative AI has flourished throughout this year and key AI giants are eyeing more exciting developments and evolution in the coming days. Behind all the success of AI and ML is the support of GPUs that gives the power of computing to train large ML models smoothly. As a result, GPUs are the rare-earth metals for artificial intelligence in this era of Generative AI. While employing parallel processing, GPU systems scale to the extent of a supercomputer and the software stack of GPUs for AI is broad and deep. Hence, it can perform fast technical calculations with greater energy efficiency than that of CPUs. This helps in delivering leading performance when it comes to AI model training and inference with gaining across applications that use accelerated computing. Learn more about how GPUs are beneficial for AI from this article published by Nvidia where they explained in detail how their GPUs give ideal machine learning performance by providing features in chips, systems, and software.

• Recently, GitHub upgraded its entire MySQL architecture from version 7.0 to MySQL 8.0. MySQL is a core part of GitHub’s infrastructure and is the relational database of choice. Starting with a single database during the start of the platform, the architecture has evolved to meet the scaling and resiliency needs of the platform including building for high availability, implementation of test automation, and partitioning of data. With MySQL v7.0 getting deprecated, an upgrade to the latest version was necessary which will provide security patches, bug fixes, and performance enhancements. Features like instant DDLs, invisible indexes, compressed bin logs, and others will benefit the org and help them try out new things. Learn more about the upgrade from this article published by the GitHub engineering team where they have discussed the journey in detail.

Now, we will head over to some of the news and articles that will be at a place of interest for developers and the tech community out there.

Good to know

• Folks at Doordash have improved their SRE experience through the standardization and improvement of Microservices caching. With the growth in the microservices architecture of the company, there is a significant growth in the volume of interservice traffic. While each team manages their own data, they expose access by means of gRPC, an open-source remote procedure call framework for building scalable for building scalable APIs. Although caching remained a go-to strategy to improve performance and reduce costs, the lack of a uniform approach to implementing it has led to complications. Here is a well-explained article from the Doordash engineering team where they discussed the streamlining of caching through a Kotlin library that offers backend developers a fast, safe, and efficient way to introduce new caches.

• As companies grow over time, the complexity of their workloads and applications increases. As a result, the challenge of debugging, optimizing, and maintaining them also increases. To tackle and amplify these challenges enter Distributed Systems where various services interact to fulfill a user’s request. Hence, monitoring the whole journey of interaction is quite complex and this is where distributed tracing comes into play. It allows the DevOps team to visualize and analyze the flow of requests as they travel across different components of an application which in return not only diagnoses and resolves issues but helps in optimizing performance and resource allocation. All these can be achieved using OpenTelemetry, which is a unified set of APIs, libraries, agents, and instrumentation to enable observability in applications that collect and observe telemetry data. This article from the engineering team at Licious discusses how their team implemented distributed tracing employing OpenTelemetry.   

• With the rise of OpenTelemetry, many companies are using the open-source tool for monitoring and collecting telemetry data like logs, traces, etc. However, collecting those data is not enough because you have to ensure best practices are in place so that the data are easy to find and correlate with other data. All these lead to good attribute naming standards. It is not just a best practice but a critical requirement. Attribute names must be consistent among every telemetry type, tool, and service so that the data remain valuable in troubleshooting and post-mortems. Without uniformity, the usefulness of the telemetry data gets reduced. Learn about the best practices to name OpenTelemetry attributes from this article published by thenewstack.

• The open-source project that caught our eye this week is UniDep. It is a command-line interface tool that provides unified conda and Pip dependencies. It can install conda, pip, and other local dependencies with a single command and can give specific support to different operating systems or architectures. Also, it can allow projects to manage dependencies using a single `requirements.yaml` file for conda and Pip dependencies. You can check out the project from its GitHub page and leave a star to show some support.

Lastly, we will take a look at some of the trending scoops that hold a special mention for the community. 

Notable FYIs

• Nvidia has been a major investor in AI startups on generative intelligence. With Open AI getting the lion’s share in the generative AI space, it has been powered by Nvidia’s H100 chips. Here is a podcast by Eric Newcomer and Chris Miller in which they discussed the history and growth of Nvidia along with the geopolitical war that is heating up with the production of chips.

• With the innovation and evolution of AI being celebrated by developers, open-source LLMs are slowly making their way to beat the AI giants in the game. With a 5x faster model than GPT4, Phind is a new player giving competition to OpenAI in the “search for developers” arena. This podcast from Latent Space talks about how Phind made it possible to build a model that is beating GPT4, the problem it solves, and how it became the number one model in the bigcode leaderboard

• If you are familiar with software engineering, you might have heard acronyms related to system design like CAP, SOLID, KISS, etc. CAP theorem and SOLID principles constitute the basics of software design and engineering. Here is a short article from ByteByteGo explaining the acronyms in detail.

• OpenTelemetry is the modern-day light in the world of observability. If you are new to observability and want to learn about it, here is a short video from the Elastic community YouTube channel that explains the vision of OpenTelemetry to provide ubiquitous observability. 

• Nowadays, data pipelines are an important component in the world of operations and automation. It is a fundamental component that helps you in the efficient managing and processing of data within a modern system. This short article from ByteByteGo gives you an overview of the different phases of pipelining in a system.  

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.