aebm.bsky.social

aebm.bsky.social

@aebm.bsky.social

24 followers41 following199 posts

aebm.bsky.social reposted

I know this is an extremely weird request, but also an extremely critical one: does anyone know of a GOOD/GREAT Canadian immigration lawyer? (yes, I can google it on my own, here I am asking for direct knowledge 😊🙏🏽). I would appreciate any repost of this message, thanks a lot, everyone! 🙏🏽🙏🏽🙏🏽

aebm.bsky.social reposted

Let's start by stating the obvious: The corporate behemoths are dictatorial, soulless hellholes that dehumanize people and do active harm to their employees. (Now tell us what you really think, Allen 😄.) Every one of them (at least the ones I've seen) is vastly ineffective in how they do 1/8

aebm.bsky.social reposted

The workplace is not a battlefield where our co-workers are the enemy. Any system of management that sets up a competitive environment will always be inferior to one based on trust and collaboration. (And I'm willing to go out on a limb with that "always." I've never seen an exception.)

aebm.bsky.social reposted

Replying to @lizthegrey.com

this is just rather depressing that the same week the second edition of the SRE book comes out detailing all of the various jobs that SREs do, even in the AI era, there are companies out arguing that you should fire all your SREs and replace them with their AI incident response bots.

aebm.bsky.social reposted

AI makes your work harder, not easier. I think of it as work density—the percent of time spent expending brain cells. The denser the work, the more exhausting it is. If you have 10 agents running at once, you have 10x the context swaps, each of which requires recovery time. 1/5

aebm.bsky.social reposted

I wish people would stop equating SRE with oncall.

Gergely Orosz@gergely.pragmaticengineer.com

When I hear about “AI SRE” startups (products hoping to automate oncall so no dev gets woken up at 2am): Putting in more work so stuff works so you don’t get paged at 2am is what we have today. Should we get more AI and less feedback systems reliability could get a lot worse

aebm.bsky.social reposted

Uno puede llegar a pensar que dos personas haciendo preguntas a la IA generativa van a obtener respuestas diferentes, pero no es así. Hoy tengo tiempo libre y voy a explicar por qué casi siempre sale 73, y esto viene condicionado por dos cosas, el consumo y el "problema de marco"

OniNoNeko@oninoneko.bsky.social

Cuando tengo que explicar que la IA no es IA, si no una máquina que reproduce lo que estadísticamente le parece más probable, pido un número del 1 al 100 Y casi siempre es 47 o 73.

Un problema que veo con usar la IA para diagnosticar algo es que es como el problema de los full body scan, te saca todo y no sabes en qué enfocarte

aebm.bsky.social reposted

“Tokenmaxxing as a business strategy was like heating your convenience store by lighting the cash in the register on fire.” - Ben Affleck ~~ He’s showing more rational business sense than half the CEOs in tech.

aebm.bsky.social reposted

Replying to @islomar.bsky.social

Some things I'd tell my younger self: - Curiosity as compass - Optimise for learning first, earn more later - Company ≠ family - Relationships last longer than jobs - Technology is a means to an end - Communication drives outcomes - Visibility creates options - Communities and mentors FTW

aebm.bsky.social reposted

🤦🏾‍♂️

ill-boo-mi@yallchemist.bsky.social

lol one of our website designers used our organization's entire monthly claude credits yesterday and now half the company doesn't know how to do their jobs. it's october 2nd.

aebm.bsky.social reposted

Most vendors will show you a demo of what their product does. Fewer explain why it's built the way it is, what they got wrong along the way, and what they'd change now with hindsight. I'm facilitating a session at O11yDay London on Oct 7 about how we actually built Honeycomb's latest features.

aebm.bsky.social reposted

¡Primer post en Bluesky! 🦋 Estrenamos cuenta hablando de MASC, una paloma juvenil con clamidia e infección digestiva. Ha pasado su primera noche, la medicación está funcionando y sus ojitos ya están mucho mejor 🥹✨ Pero todavía tiene una batalla importante por delante… 🐦💙

aebm.bsky.social reposted

Microsoft joins Amazon, Adobe, Atlassian, and Tesla in introducing AI token limits. There are now budgets set at the division level with employees being able to look up how much they’ve spent via dashboards. Additionally, their default AI model has been switched to cheaper one. Save that money!

thenextweb.comMicrosoft tells employees to stop tokenmaxxing, sets division-level AI budgetsMicrosoft EVP Jay Parikh told staff to curb AI token use, switched the default model to a cheaper one, and gave each division a formal AI spending cap

aebm.bsky.social reposted

If you really want to treat AI-generated code as a black box, your tests have to be immaculate and extensive. For example, are you testing for all of the following (not an exhaustive list and in no particular order)? All Functionality, including all corner cases. 1/6

aebm.bsky.social reposted

The point of a retrospective is to learn and adjust. Retros are not about criticism, "constructive" or otherwise. ("Constructive criticism" is usually a smokescreen for passive-aggressive behavior, in any event.) Retros are not about people—management or otherwise. 1/3

aebm.bsky.social reposted

A stakeholder who tells you they're too busy to talk about the product is not a stakeholder at all. Stakeholders have a stake in the product. The product's success or failure directly impacts them. Everybody else is just a bystander.

aebm.bsky.social reposted

A backlog is a collection of work items that you do not have the capacity to build. (If you did have the capacity, you'd just build it.) It's a black hole where work goes to die. 1/8

aebm.bsky.social reposted

Honeycomb's SRE manager wrote up their first year on the job: bad meetings they over-engineered, and the time they recused themselves from a technical call created team issues. Key takeaway: there are lots of things you could do as a manager, only some you should. www.honeycomb.io/blog/could-s...

honeycomb.ioCould vs. Should: The First Year Managing an SRE TeamAs of today, I’ve drafted this post upwards of 10 times—it’s old enough that the version I first started working on was called “Reflections on 1 Year of SRE Management” (I’m currently at 2. 5 years).

aebm.bsky.social reposted

aebm.bsky.social reposted

woohoo, one testimonial for the new "Observability Engineering", chapter 29: "Build vs Buy (vs Open Source)"! in which we take on the evergreen question of "which do you hate more, your money or your time?" honeycomb.io/book

honeycomb.ioHoneycomb’s O’Reilly Book Observability Engineering, 2nd Ed.Let authors Charity Majors, Liz Fong-Jones, George Miranda, and Austin Parker explain how to apply Observability successfully in your organization.

oh, this is great. i work on system like little baby where observability is not a major concern but this build/buy chapter is a great expression of things i've been labouring to explain to my senior leadership for weeks now. there's wisdom in this book that reaches far beyond o11y!

aebm.bsky.social reposted

alright friends and computationists, what are the signals you pick up on that make you think a team has high engineering discipline? (whatever "high" means to you) i'm not talking science or evidence. i'm talking casual conversation, heuristics, no evidence whatsoever, just radar. what trips it?

aebm.bsky.social reposted

aebm.bsky.social reposted

Las primeras copias físicas de «Menos software, más impacto» ya están llegando. 25 años haciendo software intangible y lo primero que fabrico con peso y lomo es un libro que dice: construye menos. Si ya tienes tu copia, compártela por aquí 😅

aebm.bsky.social reposted

In my normal programming work, I use AI tools in a very focused way that I've been told is "wrong," even though I'm more productive, but there we are. I'm not spewing out 20x my previous output, so maybe that's it. I'm an engineer, so when I work with an LLM, I do engineering. 1/13

aebm.bsky.social reposted

Subí esto a un Reddit de memes de El Señor de los Anillos y me complace informar que la respuesta ha dicho abrumadoramente positiva.

aebm.bsky.social reposted

Sabes ya lo que es el "efecto Carlota"? Es la tremenda ternura que esta mini gatita produce solo con mirarla. Esponjosa, suave, descarada, graciosa, guapísima y... PERFECTA. Tiene apenas 2 meses y busca hogar. #Valencia contacto 678 34 76 73 #adoptanocompres #gatos #adopcion