Matias Kreder

What an inspiring talk by Manuel Ortiz Bey, Senior Solutions Architect at Amazon Web Services (AWS), on Amazon Q Developer! Manuel showed us that Amazon Q Developer is much more than a simple "developer companion." It's a powerful tool that can empower professionals across diverse fields, from QA and finance to infrastructure administration and design. During the presentation, Manuel took us through a fascinating journey, including: - A clear introduction to the broad applications of Amazon Q Developer - An incredible live demonstration building a serverless ice cream online store, "Manis Ice Cream," deployed completely on AWS - A structured methodology for working with large language models, covering everything from requirements definition to documentation He also showcased the impressive features and integrations of Amazon Q Developer: 🔧 IDE Integration (VS Code, IntelliJ) and CLI, enabling deep interaction with the entire system ☁️ Direct availability in the AWS console and third-party integrations with Slack, GitHub, and GitLab 🛠️ MCP tools that extend functionality for AWS documentation search, architecture diagram creation, and cost analysis 💰 Advanced cost analysis and optimization capabilities, code understanding and refactoring, and security scans that identify and recommend solutions for vulnerabilities 🎨 Innovative functionalities like interpreting UI changes from screenshots, automated testing with Playwright, database querying (like DynamoDB), and complete presentation generation Also addressed the pricing model and security, highlighting that user data isn't used to retrain models and that administrators have full control over functions and usage monitoring. Someone told me (Do Rego Ezequiel) this changes everything, and after seeing this demonstration... he's absolutely right. I haven't gotten my hands on it yet to understand its limitations, but it's definitely a tool worth exploring to take your projects to the next level! #AmazonQDeveloper #AWS #Development #ArtificialIntelligence #CloudComputing #Innovation

21

5 comentarios

🚀 New Article Published: C# 14 + gRPC Customer Microservice Demo (.NET 10) While most teams still use REST for internal microservices, modern distributed systems require stronger contracts, faster payloads, and native streaming. I demonstrate how to build a Customer microservice using gRPC + Protobuf in C# 14 / .NET 10, including: ✅ Contract-first .proto design ✅ Unary RPCs (Create + Get Customer) ✅ Server Streaming (Stream Customers) ✅ Validation + clean error handling (RpcException) ✅ BenchmarkDotNet: JSON vs Protobuf performance comparison 📌 Repo: csharp14-grpc-customer-microservice-demo 👉 Read the article: https://lnkd.in/eVqP6dn7 👉 GitHub repo: https://lnkd.in/eH74fBXi Whether you build microservices, real-time systems, IoT pipelines, or AI-driven workloads in .NET, this is a must-have tool. #dotnet #csharp #grpc #microservices #softwarearchitecture #protocolbuffers #benchmarkdotnet #microsoftmvp CSharpCorner

15

I’ve seen this conversation play out dozens of times: “Should we use ECS or EKS?” “Or maybe Lambda?” The debate usually goes in circles because it starts with the wrong question. The real question is: How much responsibility does your team want to own? AWS usage patterns make this clear:   • Serverless adoption continues to grow fastest for event-driven workloads   • Managed container platforms dominate new production deployments   • Kubernetes remains critical, but mostly where platform maturity exists AWS documents this directly in its compute decision guides: https://lnkd.in/ePdtJGDg Here’s the framework I would use: Choose Lambda when:   • Event-driven architecture   • Spiky or unpredictable traffic   • Minimal operational ownership Choose ECS when:   • You want containers without Kubernetes complexity   • Speed and standardization matter   • Opinionated defaults are a feature Choose EKS when:   • You need ecosystem leverage   • Platform teams exist   • You intentionally accept operational tradeoffs None of these is “better.” Each has a cost of responsibility. Good architecture starts with intent, not preference. Which of these decisions do teams struggle with the most in practice? , #AWS #Containers #Kubernetes #Serverless #CloudEngineering #ArchitectureDecisions

30

Chile is leading the way towards an open source LLM for Latin America. Pretraining data comes from 21 countries and they collectively contributed more than 2.5M documents. What is interesting to me is that they also suggest a way to help improve Latam-GPT by participating in conversations that will be used to train the model. They are asking this for a better model development. https://www.latamgpt.org/

9

𝐄𝐱𝐩𝐥𝐨𝐫𝐢𝐧𝐠 𝐂𝐨𝐬𝐭-𝐄𝐟𝐟𝐢𝐜𝐢𝐞𝐧𝐭 𝐋𝐋𝐌 𝐈𝐧𝐟𝐞𝐫𝐞𝐧𝐜𝐞 𝐰𝐢𝐭𝐡 𝐀𝐖𝐒 𝐍𝐞𝐮𝐫𝐨𝐧 After deep-diving into #AWS Neuron accelerators, I've published two Llama models to help teams evaluate cost-optimized inference strategies: 🔹 Llama 3.2 1B FP8-Neuron 🔹 Llama 3.3 70B FP8-Instruct-Neuron 𝐊𝐞𝐲 𝐇𝐢𝐠𝐡𝐥𝐢𝐠𝐡𝐭𝐬: • f8_e4m3 quantization optimized for #Inferentia2/#Trainium Neuron cores. • Quantized using NXD (#Neuronx Distributed Inference) library with vLLM inference support. • Pre-compiled artifacts included for fast deployment and startup times. • Benchmarking script included to test for varied input/ output lengths and concurrency. • The 1B model achieves ~850 tokens/sec on a single Neuron core[2 chips], scaling to ~2.1K tokens/sec with tensor parallelism of 8 [4 cores] while the 70B achieves a modest throughput of 600 tokens/sec on 12 cores. For teams balancing performance and cost, adding Inf2 instances to your LLM inference fleet can significantly reduce operational expenses. Are you evaluating if Neuron is the right fit, Do try out the below smaller than small quantized model, it contains instructions and precompiled artifacts for a quick evaluation. 𝐌𝐨𝐝𝐞𝐥 𝐥𝐢𝐧𝐤: https://lnkd.in/dCRMB-Rs #AWS #AWSNeuron #AI #CostOptimization #vLLM

18

❤️‍🔥 NEW EPISODE! This week the entire Ethereum ecosystem celebrates the 10th mainnet anniversary 🎉 Decentralized Voices is part of the Ethereum community, and I'm happy to feature Mayra C., People Operations Specialist from Ethereum Foundation. 🎯 In this episode, Mayra and I discuss hiring culture and employer branding in Web3, red flags for candidates during the hiring process, and the most important skills needed to get hired in 2025. Key takeaways: 🔸 Values and culture at the Ethereum Foundation 🔸 The current state of hiring in Web3 🔸 Identifying red flags in job offers 🔸 Vibes, followers, or skills: what real talent looks like in Web3 If you're a hiring manager in Web3 or a job seeker, this conversation will be insightful for you! Tune in on YouTube, Spotify, and Apple Podcasts ⤵️

23

3 comentarios

There have been lots of interesting LLM releases last week. My favorite was actually the Olmo 3 release. Olmo models are always a highlight since they are fully transparent (including training methods and datasets) and come with very detailed technical reports. I am sure I'll talk more about the interesting training-related aspects from that 100-pager in the upcoming days and weeks. In the meantime, here's the side-by-side architecture comparison with Qwen3: 1) As we can see, the Olmo 3 architecture is relatively similar to Qwen3. However, it's worth noting that this is essentially likely inspired by the Olmo 2 predecessor, not Qwen3. 2) Similar to Olmo 2, Olmo 3 still uses a post-norm flavor instead of pre-norm, as they found in the Olmo 2 paper that it stabilizes the training. 3) Interestingly, the 7B model still uses multi-head attention similar to Olmo 2. However, to make things more efficient and reduce the KV cache size, they now use sliding-window attention (e.g., similar to Gemma 3). Next, the 32B model (the figure is not shown here due to space reasons, but you can find it in my "The Big LLM Architecture Comparison" article or my Olmo 3 from-scratch notebook): 4) Overall, it's the same architecture but just scaled up. Also, the proportions (e.g., going from the input to the intermediate size in the feed-forward layer, and so on) roughly match the ones in Qwen3. 5) My guess is the architecture was initially somewhat smaller than Qwen3 due to the smaller vocabulary, and they then scaled up the intermediate size expansion from 5x in Qwen3 to 5.4 in Olmo 3 to have a 32B model for a direct comparison. 6) Also, note that the 32B model (finally!) uses grouped query attention. And yes, I also did a from-scratch implementation. It was still a lot of work, but since I had already implemented Qwen3 from scratch, as well as Gemma 3 (for the sliding-window attention component), it wasn't too bad! If you are a coder, looking at the from-scratch implementation, it's probably the easiest way to understand the architecture: https://lnkd.in/gQgxtVUu

1558

34 comentarios

Kubernetes is not a strategy.  I've seen too many organizations adopt K8s without clear business outcomes in mind, creating spectacular technical achievements that solved no actual problems.  One LatAm company spent 8 months on Kubernetes migration only to realize their monolith was actually perfect for their needs.  Tools serve strategy, not the other way around. Start with "why," not "how". 🔍 Looking for the details? Head to the comments for the link. #Kubernetes #CloudNative #Alignment

23

2 comentarios

Struggling to maintain visibility across all your microservices and distributed projects? As architectures become more complex, ensuring consistent code quality and security across the board is a major challenge. That's why I'm excited to that Sonar now has an integration with Port.io! 🎉 By bringing #SonarQube's deep code intelligence directly into Port's software catalog, we’re giving engineering leaders, platform teams, and developers a single pane of glass for engineering excellence. With this integration, your teams can: 🔹 Get a centralized view of code quality and security metrics. 🔹 Monitor SonarQube Quality Gates for every service directly within the developer portal. 🔹 Reduce context switching and help developers prioritize remediation tasks effectively. Ready to build a centralized hub for engineering excellence? Learn more in our latest blog post. #PlatformEngineering #DeveloperExperience #CodeQuality #CodeSecurity #DevOps

5

1 comentario