The DevOps Letters Weekly #24
What is and isn’t a CI part of CI/CD aka Continues Integration and Continues Deployment 🚀🌱
Hi everyone,
Welcome to the 24th edition of The DevOps Letters Weekly! In this issue, I have curated a collection of insightful articles covering a wide range of topics in the world of technology. From generative AI and its risks to the challenges and opportunities of edge computing, there's something for everyone.
🤖 How does grammar checking work in tools that we use daily?
☣️ AI Safety Summit introduced by the UK government
📈 LLM apps monetization
🚀 How to start architecting your LLM apps
🗺️ What is and isn’t a CI part of CI/CD aka Continues Integration and Continues Deployment
🏦 Regulations in FinTech
Let's dive into the content. If you have any comments or ideas, please feel free to write them down in the comment section. Also, click the 👍 and 🔔 button if you find this content useful and want to stay up to date with the latest technology.
Happy reading!
AI
How does grammar checking work in tools that we use daily?
Grammar checking at Google Search scale
The Google Search team shows us how the process of translation and spellcheck works in one of the most commonly used search engines in the world.
Many people with questions about grammar turn to Google Search for guidance. While existing features, such as “Did you mean”, already handle simple typo corrections, more complex grammatical error correction (GEC) is beyond their scope. What makes the development of new Google Search features challenging is that they must have high precision and recall while outputting results quickly.
Under the hood, they use custom LLM which predicts user intentions and enforces a spellcheck engine.
Looking at the above diagram it looks like this is a quite complicated model training process. I wonder how much this process costs and what infrastructure is used from a DevOps perspective.
But finally as a result we have a pretty well and accurate help in our daily life 🤔.
AI Safety Summit introduced by the UK government
AI Safety Summit: What to expect as global leaders eye AI regulation
AI, LLM, and AGI are hot topics this year but every new technology especially such advanced tech needs some kind of regulation. The same movement we can observe in the previous year about Blockchain.
The AI Safety Summit, convened by the UK government, is the latest in a series of regional and global political initiatives to shape the role AI will play in society.
The UK government has started work on such regulations, first step is an AI Safety Summit led by a prime minister.
Frontier AI, as defined by the UK government, refers to highly capable general-purpose AI models that can perform a wide variety of tasks at a level that meets or exceeds the most powerful technologies available today.
As frontier AI we can treat all models like GPT 35, GPT 4, and so on.
Architecture Weekly
Oskar Dudycz mentions in his latest issue the topic of LLMs which is hot right now. What is your opinion about the monetization of this technology?
It’s interesting to see if this business model is viable. Maintaining and running Large Language Models costs a lot. Generative AI has many use cases but is also limited in the scope of usage. Those tools will need to be more cost-effective and might not be as global usage as predicted.
At the current stage of adoption and from my perspective there is still much to do in the area of knowledge spread and education of companies on how they could adopt LLMs. Also, tooling needs to be more mature I’m talking about this from the perspective of my recent works with a Semantic Kernel.
How to start architecting your LLM apps
The architecture of today's LLM applications
One of the biggest problems when architecting LLMs is the fact that their responses are not probabilistic. This gives us some interesting elements to discover. In a recent GitHub article, Alireza Goudarzi and Albert Ziegler try to explain how we can split this problem into smaller pieces.
NASA and IBM recently open sourced the largest geospatial AI model to increase access to NASA earth science data. The hope is to accelerate discovery and understanding of climate effects.
There are also a lot of ongoing projects in the area of LLM apps for me most interesting is what NASA is doing 👆
DevOps
What is and isn’t a CI part of CI/CD aka Continues Integration and Continues Deployment
Continuous Integration: That’s Not What They Meant • Clare Sudbery • YOW! 2023
Probably many times during the recruitment process you are asked about basic stuff like CI/CD or you wonder if properly treat these steps. This video explains this interesting topic.
Regulations in FinTech
Beyond DevSecOps: Why fintech companies need to consider DevSecRegOps
Why is it important to shift left and think about regulatory requirements during the implementation of the DevOps process
Automating compliance and regulatory checks is the most effective way to ensure compliance standards are met. One way to achieve this is to build regulatory checks into your CI/CD pipeline to ensure consistent compliance with auditable trails. Ideally, ensuring these compliance checklists trigger a failure close to the beginning of the SDLC ensures you don’t get to the end and realize you’re not compliant.