Luc Juggery

A 𝘀𝗰𝗮𝗹𝗮𝗯𝗹𝗲 and 𝘀𝗲𝗰𝘂𝗿𝗲 system built in 𝟯 𝗺𝗼𝗻𝘁𝗵𝘀. Two developers. No message queues. No external caches. Now powering 2,500 secure, real-time messages per second for over 𝟭𝗠+ 𝗵𝗲𝗮𝗹𝘁𝗵𝗰𝗮𝗿𝗲 𝗽𝗿𝗼𝗳𝗲𝘀𝘀𝗶𝗼𝗻𝗮𝗹𝘀 with 𝟵𝟵.𝟵𝟵𝟵𝟵% availability. Akka gave Doctolib the foundation to deliver: ✓ Seamless clustering for instant scale ✓ Low latency via sharding and pub/sub messaging ✓ Redundancy across multiple servers ✓ Simple deployment and ongoing management “It just works!” -Jasper Aarts ⬇️ Read the story #Innovation #Technology #Engineering #Java #DistributedComputing

32

4 commentaires

Did you miss this post by Fabien Potencier ? https://lnkd.in/erfHp4KG Great and insightful view of the new Symfony #AI initiative 🚀 Two items seem particularly relevant : 🔧 MCP SDK: An implementation of the Model Context Protocol, enabling your applications to communicate with AI systems using the emerging industry standard. ⚙️ MCP Bundle: Allows your Symfony applications to act as MCP servers or clients, opening up new possibilities for AI integration and tool creation. Whether you like it or not, MCP is going to play a big part in the coming years !

8

data.gouv.fr just released an experimental MCP server, enabling AI agents to query french open data in natural language, no custom API integration needed. An open source initiative with a clear ambition around digital sovereignty, while honestly acknowledging LLM limitations (hallucinations, reliability). MCP is steadily becoming the standard for interoperability between AI agents and real-world systems. Seeing the public sector embrace it is a strong signal. #MCP #AgenticAI #openData Official announcement: https://lnkd.in/ejKgp3nM

21

1 commentaire

Java 26 is coming this March with several interesting updates and features. Loïc Mathieu breaks down what's landing in this release, including: • Stable releases of Vector API and Scoped Values • Preview features like Module Import Declarations • Incubator updates for Primitive Types in Patterns and Code Reflection API • Performance improvements and bug fixes Whether you're planning to upgrade or just want to stay current with Java's evolution, this article covers the key changes you need to know. Read the full breakdown here: https://lnkd.in/euurYNBv #Java #Java26 #OpenJDK #Programming

21

I wrote the Claude Code guide I wish had existed when I started. 11 chapters. Zero-to-hero. • Installation → orchestrating agent teams • CLAUDE.md hierarchy → context engineering • Custom skills, MCP servers, hooks, worktrees • Enterprise security and sandboxing After 17 years of building production systems, I've found that the best documentation comes from the trenches. #AIEngineering #ClaudeCode #DeveloperTools https://lnkd.in/g-6cz-et

18

Launching today: Kubernetes, Uncontained by Flox! 🎉 We’re rethinking how teams build and deploy on Kubernetes. With Flox, you can now run the same immutable, declarative environment across local dev → CI → production Kubernetes clusters — on x86 and ARM — with: 🟣 Faster deployments 🟣 Built-in reproducibility 🟣 SBOMs-by-default and atomic rollbacks Who is it for? 🟣 Platform & SRE teams: Simplify ops, skip image rebuilds, roll back instantly. 🟣 AI/ML engineers: Define models as hash-pinned environments, reuse dependencies across pods. 🟣 Security teams: Reduce your attack surface, get verifiable SBOMs and provenance automatically. 🟣 Developers: Develop and deploy in the exact same environment with zero drift, faster iteration. We’re unveiling it this week at #KubeCon North America, Booth #1850 in Atlanta. ✅ Live demos ✅ Chats with the Flox engineering team ✅ Special guest Kelsey Hightower! Come see “Kubernetes, Uncontained” in action and try for yourself today! 👉 https://buff.ly/Gxqjlor Check the comments for the technical walkthrough, demo video, and press release!

54

2 commentaires

🚀 Frontier Friday #41 | Claude Code Agent Teams    Anthropic’s Claude Code is taking a major step forward with the introduction of #AgentTeams. Instead of a single assistant, Claude Code can now coordinate multiple specialised agents — planning, implementing, reviewing, validating and much more — working together like a real delivery team.   AI is evolving at record-breaking speed and it can be hard to keep up, let alone fully understand. Check out our latest insights, broken down for an easy lunchtime read.   Today we cover: 🔹 How Agent Teams work 🔹 Where they add the most value 🔹 Enterprise considerations and risks 🔹 What this means for the future of DevOps   We stay curious so you can stay informed 👍   #FrontierFriday #ArtificialIntelligence #DevOps #ClaudeCode #AgenticAI

20

I've been running Claude Code to build my new project — but until recently, I couldn't tell what a feature costs. Full TDD cycle, GitHub issue creation, PR creation, all handled autonomously. One thing I'm missing: how much does each feature actually cost? Claude Code emits OpenTelemetry metrics (tokens, USD cost), which is great. But they're tagged with session UUIDs, not anything meaningful like a feature name or ticket ID. You can add RESOURCES, but that's static information — you can obviously update it every time you start a new feature, but that's not a great solution. So I built a workaround: OTel collector → Prometheus → Grafana, with textfile metrics mapping session IDs to feature IDs (via NodeExporter). It works — until you run concurrent sessions. One more issue: the OTel Collector expires metric series after 5 minutes of inactivity (by default) — so if Claude Code goes idle mid-session, the next conversation starts a fresh accumulation from zero. The workaround is to increase expiration time but this is not ideal solution. Here's a real example from a recent PR: Cost: $1.49 Tokens: 3,755 input / 30,300 output / 3M cache-read / 140K cache-write That's a full feature — tests, implementation, PR. The fix is straightforward: when you set a custom session title in Claude Code, emit it as an OTLP resource attribute. One label. The Grafana query becomes: sum by (feature_id) (claude_code_cost_usage_USD_total * on(session_id) group_left(feature_id) claude_session_feature) Filed a feature request with Anthropic: https://lnkd.in/eW_4phzy It's sitting alongside 5,000+ open issues, so I'm not holding my breath :) If you're using AI coding agents in any serious capacity, per-task cost visibility is a necessity. How are you approaching this? There may be simpler solutions I'm missing.  #ClaudeCode #OpenTelemetry #AIEngineering #DevTools #EngineeringLeadership #Observability

3

A good read for anyone thinking beyond single-agent LLM workflows. It doesn’t just talk about the architecture — it also dives into the harder but equally important aspects: evaluation & reliability, production readiness, engineering challenges: https://lnkd.in/eZ7kpUxD

4

You want to build something. You open your editor, design the perfect architecture, set up your CI/CD, configure Kubernetes, and plug in a global CDN. You tell yourself you’re “building for scale.” But let’s pause for a moment — who are you scaling for? Do you have users waiting for this? Or are you optimizing for a problem that doesn’t exist yet? Most startups don’t fail because of bad code. They fail because founders build for scale before they build for someone. As engineers, we love structure. We love solving problems elegantly. But in startups, elegance is a luxury — relevance is survival. Every hour spent perfecting code before validating demand is time spent deepening the illusion of progress. We think the next feature will bring users. It won’t. We think the better UI will fix engagement. It won’t. We think better architecture will attract growth. It won’t. Because startups don’t die from poor engineering — they die from silence.

2