Peter Friese

One of the more requested examples: how to use the Code Interpreter SDK inside a Next.js app with streaming enabled and using the Vercel's AI SDK

10

Here are the slides for my GopherCon.EU talk: How to Win Frames and Influence Pointers https://lnkd.in/d7byRQfu #gopherconEU #golang

71

4 Kommentare

The benefits of AI code assistants are still being evaluated, but here at Deutsche Telekom we are confident that our AI strategy to continually assess COTS products, as well as developing our own products and data foundation is the right approach to yield benefits in the future. Have a look at the post below from one of our leading developers on the topic It’s exciting to think about the capabilities of the product in only a few years from now; perhaps covering all aspects of the SDLC and integration to user stories, more context in relation to existing code repos throughout the enterprise (particularly useful for very large enterprises), models optimized for dialects of the languages (GoogleSQL vs PostgreSQL etc)….to think of a few examples Regardless, the developer experience will be greatly enhanced in the next few years and at Deutsche Telekom we’ve got a front row seat! 👏

15

1 Kommentar

“In the search for solutions, do not be content with the surface; dive deep into the depths where non-obvious errors and situations await discovery.” Building on our previous conversation on asynchronous error handling, I explore ways to handle errors involving workflows, the role of idempotency, and how to connect all together to keep your support team sane! https://lnkd.in/d-zFD5dQ

18

🌟 Hey NET devs, time to share some exciting news! 📝 Are you interested in learning how to use Ollama and Phi-3 with C# for free? Our new repository at https://lnkd.in/g-UBe8Eq allows you to set up a test environment in seconds! 📎 This environment provides access to a Phi-3 model that you can experiment with using sample C# projects. It's a fantastic opportunity to enhance your skills in AI and machine learning. 🔗 Check it out and let's us know what other samples you want in the repo! #CSharp #AI #NET

29

What is the ideal number of mobbers? A guideline: ”If you feel you are contributing or learning, you might as well stay with the team. Otherwise, you can temporarily move off the team and work on something alone or with a pair. The basic idea is known as the 'law of two feet' in the Open Space Un-Conference community.” ― Woody Zuill https://lnkd.in/gptmBTE #MobProgramming

33

2 Kommentare

How many of your supposedly internal-only backends are serving critical end-user traffic? Probably more than you think! Most frontend services have many transitive dependencies, and understanding these dependencies becomes increasingly difficult as your system grows in scale and complexity. As a result, it becomes exceptionally tricky to ensure that internal-only backends are fully isolated from end users. Google uses a common infrastructure platform for almost all of its services, internal and external; this provides for a message-passing mechanism common to all of these services. One outcome is that systems engineers can build accurate dependency maps of services; and that we can ensure new services only depend on other services at the same level or higher in the dependency stack. Risky dependencies — in this case, dependencies which place internal-only backends on the critical path for end users — can be a source of major production outages, because the internal-only backends do not generally have necessary reliability or safety guarantees. For example, you might have an externally-visible service that requires high levels of availability and performance, and this system has an indirect dependency on a backend with no availability or performance SLOs. At Google, we found success preventing major and huge outages by clustering services into categories with distinct properties, clear perimeters, and limited blast surfaces. The properties will depend on the guarantees we want to enforce. In this article, we use Google Maps to demonstrate how risky dependencies proliferate in complex systems, and then showcase how clustering Google Maps services into just two categories ("Internal" and "External") can prevent major and huge serving outages. We also outline how OpenTelemetry is used to identify violations of this clustering, which can be fixed before the violations cause a major serving outage. This article is based on a talk at SREcon EMEA, which you can see here: https://lnkd.in/e54b6G6p.

1

LLMs are trained to a past date. They are unaware of the enterprise's internal data, documents, current affairs etc. LLMs also hallucinate (make up believable answers that are not facts). Does that mean LLMs are not useful to enterprise use cases? Fortunately, the RAG (Retrieval-Augmented Generation) framework enables you to retrieve the relevant data for a given prompt from your dataset, augment that as a context in your prompt to LLM and finally generate responses that are more accurate, less prone to hallucination and relevant to enterprise use cases. Langchain is one such popular framework to quickly set up RAG for your use case. I have personally tried and found it very easy to create an RAG application with very nominal effort. For retrieval, you would need to clean, split your data/documents and then embed using an embedding model (I tried OpenAI and llama2) and finally index your data in a vector database of your choice (I tried FAISS and Weaviate). These vector databases give functionality to find semantically matching chunks for a given embedded query. Given the increasing popularity of vector databases, legacy popular databases like Mongo and Postgres (with a plugin) have also started supporting vector search. Interestingly a lot of new companies started offering new Vector DB solutions. This reminds me of the quote "During a gold rush, sell shovels" 😊 #AI #LLM #RAG #Langchain #VectorDB #MongoDB #Postgres #OpenAI #llama2

36

2 Kommentare

Feedback: Not just a cool effect for punk music in the 80's! I rely heavily on #feedback in so many things in my life: Is my heart rate right? Is my #code passing the tests? Is my bicep getting bigger?(yes, obviously) Even as I type this the browser gives me feedback that obviously was spelled wrong... But what happens when feedback isn't obvious. The other morning while running my watch wasn't tracking my heart rate properly. This was fine for me because I could feel by my breath and pace that I was still meeting my goals. Sometimes though, thinking someone is aware of their performance isn't enough. As a high school administrator I remember once walking into a classroom and seeing the front tow rows engaged in what the teacher was doing and the back three were on their phones, chatting, sleeping, doodling, basically all the things I used to do when I was in high school (my phone had a really long cord.) In my mind I thought it was obvious that the #teacher wasn't meeting the standard of getting everyone engaged and that the teacher would know that the conversations and laughter from the back of the class was distracting to those that wanted to understand the lesson. To my surprise when I asked him how he thought the observation went he was pleased that he got 20% of the students engaged and thought it went well. When I reminded him that this was well below the standard he was upset and a little angry. And to be fair it is difficult in the modern day to keep kids engaged. Also though, once he was made completely aware that he wasn't meeting the standard he sought out help on things he could do to increase engagement and work on some classroom management strategies as well. I could have just left the room, assumed that he knew his performance wasn't good, I mean, it was written down that 80% of the class would be engaged 80% of the time... But, sometimes overt tough conversations have to happen to get folks to realize that they need to improve to meet standards. In my favorite podcast, "Coaching for Leaders" with Dave Stachowiak it was mentioned that criticism is investment. If my manager gives me a critique, even if or maybe especially if it's not positive I feel like he is investing in my future. He is tough on me because he wants me to be the best me that I can. He could sit back and assume I know if I'm not meeting the standard, and he would continue to get the same performance from me. But he knows I want to do a good job so he is overt when I could do better. He does deliver the message with kindness and encouragement of course which made it much more palatable, but the message still came accross. Bottom line, if you want improvement in #performance don't assume anything, provide the tough feedback and form a pathway towards #growth! Thanks for reading y'all and remember the ABCs of coding Always Be Coding and sometimes Always Be Critical (and kind).

14

2 Kommentare

Having to choose between a new technology or an existing one is hard. More often than not we see diametrical opposite positions and unproductive discussions that do not bring the real benefits and trade-offs. Using the serverless x kubernetes context, I wrote about what we should focus on and proposed one framework to guide your discussions on this topic. Check more here: https://lnkd.in/eWjKne6i

26

2 Kommentare

Stranger Danger: Code Like Your Mom Taught You "The Principle of Least Knowledge guides us to reduce the interactions between objects to just a few close "friends." The principle is usually stated as: 'talk only to your immediate friends' But what does this mean in real terms? It means when are designing a system, for any object, be careful of the number of classes it interacts with and also how it comes to interact with those classes. This principle prevents us from creating designs that have a large number of classes coupled together so that changes in one part of the system cascade to other parts. When you build a lot of dependencies between many classes, you are building a fragile system that will be costly to maintain and complex for others to understand. ... Q: There is another principle called the Law of Demeter; how are they related? A: The two are one and the same and you'll encounter these terms being used interchangeably. We prefer to use the Principle of Least Knowledge for a couple of reasons: (1) the name is more intuitive and (2) the use of the word "Law" implies we always have to apply this principle. In fact, no principle is a law, all principles should be used when and where they are helpful. All design involves tradeoffs (abstractions versus speed, space versus time, and so on) and while principles provide guidance, all factors should be taken into account before applying them." ― Eric Freeman & Elisabeth Robson, Head First Design Patterns: Building Extensible and Maintainable Object-Oriented Software https://lnkd.in/dPYV-8jF #refactoring #extremeprogramming #modernagile #agility #tdd #testdrivendevelopment #legacycode #continuousimprovement #maintainability #softwarequality #codequality #softwaredesign #lean #oop #solid #designpatterns

13

In my next webinar "Playwright in an hour" I'll introduce you to the main features of Playwright, what it can do for you, and what you should know going into the theater. Europe: https://lnkd.in/eHwxvyhf US: https://lnkd.in/e8C2dvsK #Testing #Playwright #WebTesting #WebAutomation

2

In 2009, Aza Raskin wrote about browsers helping you log in to web apps https://lnkd.in/eBD534Xb In 2011, I blogged about it in https://lnkd.in/ecQwwG39 (it was actually my second blogpost ever) and acknowledged that: > The hard part would be to get both FireFox and Chrome to say “Yes, let’s fix this blatant UX caveat.” Fast forward another 13 years, this dream is actually coming true thanks to FedCM! <3 Read all about it here on the IndieWeb Wiki: https://lnkd.in/eHP-bnBV You can self-host your own identity provider, register it in Chrome Canary, and then use it at websites, blogs, unhosted apps, solid apps, etc that accept self-hosted identities. This is so amazing... Thanks Sam Goto, Tim Cappali, Aaron Parecki, Jan Schill, Theo and many others for making this happen!

12

2 Kommentare

🔧 Ever felt the need for a robust, scalable, and open-source CMS to create professional websites or applications? Look no further! I recently discovered Contao, a powerful open-source Content Management System (CMS) that has made my life so much easier. Here's what I love about it: ⚡️ Key Features & Benefits: - **User-friendly:** Intuitive interface for both developers and editors - **Flexible:** Extensible with various themes and plugins - **Scalable:** Suitable for small projects and large-scale applications - **Secure:** Regular updates and strong community support - **Multilingual:** Supports multiple languages out-of-the-box 🌍 Personal Experience & Value: I've used Contao for various projects and found it to be the perfect blend of power, flexibility, and ease-of-use. With its modular structure and clean codebase, Contao has proven to be a reliable and efficient choice for my web projects. #Contao #CMS #OpenSource #TechCommunity #WebDevelopment

4

We live in a time when we should no longer close our minds to artificial intelligence. For this reason, after all these years, I took a look at GitHub Copilot and tried it out in my hobbyist project Life.NET. You can read about how it went, what I wanted and whether everything worked on Dr. Windows. #dotnet #csharp #blazor #copilot #ai #github #development #developer

3

AI Designs in 60 Minutes! But is it a Design Dream Team... or Disaster? Forget months of brainstorming - AI promises product concepts in a single hour. Sounds too good to be true, right? We put it to the test. AI generated exciting ideas in mere minutes, showcasing its potential. However, a closer look revealed limitations. AI excels at sparking creativity and visualizing concepts quickly, but it lacks the human touch needed for a polished, user-focused design. **Ready to explore the future of design?** #AI #productdesign #futureofdesign

2

🔍 The embarrassing truth about our (internal) feature documentation! 🤯 As the lead behind the points-based rewards system in Germany (Pluspunkte) and across Europe, I often face questions from business teams about our system's features. Being in this space for quite a long time, I typically have a good grasp of the features. …and then they ask for details 😱 Now we have to look them up in the source code as our feature documentation is - …limited. It is rather an architecture documentation and a list of features, and we avoided details on purpose as documentation and implementation typically get out of sync over time. ...and the documentation is just for us developers, isn’t it??? 🤦 This not only takes quite a long time and hence is not really efficient - it also risks our credibility with other teams. But there is a solution in the pipeline! Our system has a huge suite of acceptance tests which get executed as part of the unit tests (yes, that fast!). These tests are written in Gherkin notation, so “given - when - then” and are based on the Java BDD library JGiven. They describe the features in great details and with examples and offer a promising path. By converting JGiven's JSON report into a markdown file structure through a custom tool, we can seamlessly feed it into our Docusaurus documentation project. Exciting, right? 🤔 Do you have a better idea on how to create detailed feature documentation that doesn’t get out of sync over time? Please share your thoughts in the comments. #eBayInside, #AgileDevelopment, #AcceptanceTests, #SoftwareDocumentation

1