Marcos F. Lobo

👋🏼 Marcos here and I love these topics: ✳️ Software Engineering ✳️ Technical Leadership ✳️ Technical Growth I help you to be better on those topics here down 👇 newsletter.optimistengineer.com All with a bit of optimism 🌱 #softwareengineering #leadership #growth #optimistengineer

New issue of my newsletter The Optimist Engineer is out! newsletter.optimistengineer.com/p/the-experi...

I find this distinction really important ainativecompass.substack.com/p/subagents-...

Was great be speaking at the React Advanced/TechLead conference in London, yesterday! My talk was titled: How to Build AI-Native Engineering Teams.

✨  Django Girls Barcelona at #PyConES2026! 🐍💖 👩‍💻 Free coding workshop for women (apply by Oct 25th). 💡 We're looking for Django/Python mentors! 👉 Apply or join as a mentor: https://f.mtr.cool/9yz5p5wbe8 #PyConES #Python

Status updates. In my experience, the blockers are solved/tackled on the fly as they appear by having a chat or an extra call.

Proper 2 pane tablet support, that's the headline feature for me. Bluesky app is very poor, it's a web app that doesn't scale properly.

Folks, do you see this? The internal tools we are creating nowadays in the companies are meant to be used through AI agents, not through browsers.

Pues ya estamos montando el equipo. Si te interesa, ya puedes ver las ofertas que tenemos publicadas. Staff y Senior AI Platform Engineers -> technosylva.bamboohr.com/careers/234 technosylva.bamboohr.com/careers/233 Si vas a estar en la #TRG26 , te lo cuento en persona

CLEAR-PC cumple su primer año con avances para mejorar la información y las decisiones sobre el cáncer de próstata. @clearpc-project.bsky.social @astursalud.bsky.social ficyt.es/blog/clear-p...

The latest issue of my newsletter is here! 👇🏽 newsletter.optimistengineer.com/p/the-ai-sta...

@theroundedman.bsky.social mira que chulada hizo esta chica juliadiez.substack.com/p/i-taught-1...

Si estás buscando aprender acerca de #AINative en #español, esto te interesa ainativebrujula.substack.com

It seems Opus 5.5 is still on the top of the wave. For how long? llm-stats.com

When your team is playing, Dash drops everything: a full-screen live board takes over (football, MLB, NBA, NFL, NHL, tennis, F1), and when you score you get a full-screen celebration 🎉 Rivals score? Just a small toast. buff.ly/jf4xfPe

Wish I had more user feedback for nubecita. https://nubecita.app

Most likely, for your first MCP server you will not need a RAG nor something sophisticated... But just expose a couple of tools to describe your data and another tool to query it. It doesn't mean your MCP server is not challenging though!

I am #lookingforwork Do you need someone who’s good with AI language models, likes speaking to politicians about tangible risk rather than sci fi risk, or a defense geek who likes flying robots? Well I’m your guy! My signal is NoahW.03 I can legally work in both the United States and Canada.

This is Pele, my new rendering library for diagrams and charts, compatible with the latest Mermaid syntax. - 100-500× faster than mermaid.js - 6× smaller bundle - No dependencies - Easy to theme - Responsive to device width - Open source (MIT) Try it at pele.run

The key here is to agree ahead of time what resourcing you will set aside to deal with technical debt, and then sticking to it. Sometimes business needs require moving fast, and sometimes shipping good enough now is the right choice. But if you don't protect your time, you won't get back to it.

In my case, we solve technical debt at the same time as we implement new features for the product. This has 2 benefits: 1. Helps to solve tech debt in small chunks 2. You actually solve tech debt

🛎 ¡Este viernes tenemos el evento de octubre! GitHub Dev Days + Fishwbol, un evento doble con una charla ("Del prompt al agente" by Diego Zapico) y una dinámica más que interesante. 🗓️ 16 Octubre · 18:00h 📍 Talud de la Ería · Oviedo Apúntate: buff.ly/c4Y9uFf

This is good stuff newsletter.pragmaticengineer.com/p/building-r...

Interesting learnings about error handling in MCP servers 👇🏽

👇🏽

When running local AI agents on terminal tasks, how do you handle tool timeouts? Loose timeouts usually leave zombie processes holding ports when an interactive prompt blocks. I ended up wrapping commands in process groups with a strict sigkill grace window. #devtools #automation

Four failure enums are enough if each changes the next permitted action: retry, revise input, refresh authority, or stop. Otherwise normalization only makes ambiguity shorter. Do you keep the original error behind the enum for audit, or discard it after classification?

Interesting technique from experience 👇🏽

Back to basics always work, more over in times of AI generating our code newsletter.optimistengineer.com/p/symptoms-o...

⚡ VALHALLA, NOW! After years of waiting, Value Types could finally land in JDK 28 as a preview. Join Nicolai Parlog @nipafx.dev to discover what they can do, what they can't, and where Valhalla goes next. 📅 Oct 21, 19:00 📍 Oracle Madrid 🔗 www.meetup.com/es-ES/madrid... #Java #JVM #Valhalla

Error shapes and pagination blow up first for me. Most MCP servers dump a raw 50KB JSON blob or a generic 500 string back into context, and the model burns three turns guessing whether to retry or change parameters.

The most difficult part of creating MCP server is the definition of the tools to be exposed Any other? ainativenewsletter.com

Margaret Hamilton, computing pioneer who led software development for the Apollo program, dies at 90 news.mit.edu/2026/margare...

Your weekly issue is here again! optimistengineer.substack.com/p/the-a-side...

New episode! Roxy Rodriguez-Becker on going from middle school math teacher to developer community manager: studying at 4AM, treating networking as a muscle you build, and shipping a tube feed tracker that real families rely on. 🎙️

I can see that generative AI systems, like Claude Code, are helping the Makers culture great again... And that's good.

It's funny because it's true newsletter.pragmaticengineer.com/p/the-pulse-...