Systems Pimp 🎩🦯🕶️

Exploiting bad design with its own rules.
Turning broken systems against themselves.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@dev_maims Ai can do a lot of the more advanced things, but it never does it on its own. Tastes, preferences, and vision are in high demand. Having a great w2 job feels like opportunity cost now.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@ankkala When doing programmatic or automatic I like to stat files and build a pandas dataframe and save to csv or sqlite. Then you can answer any question you have. easily.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@ankkala I use du for that

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@ankkala I use du -hd1 and going from memory feel like it gets dot dirs

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@lafaiel I really like my snapdragon laptop.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@usr_bin_roygbiv @usr_bin_roygbiv think bigger. Make a recruiter llc and get kick backs on placements.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@euboid I think you are making a tactical error. The story of what we did and why, is helpful to your team and the agents.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@tantaman Your factory doesn't do these things? Because you are describing quality gates.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@Olking07 The public library and a TRS80 is how I learned.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@southpolesteve CI/CD and feature flags. Flagd is pretty solid.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@0xCodio I put the model in control of context a few months ago and recently started stressing it live. Gemma4:26b with 2 days 18 hours as a single session.
reasoning is an inverted scale based on context size floored at 25% used.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@52_65_64 I had to use coding style variables, but I always did better at word problems in math. Some of us need the why context.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@madras_bertha It would seem logical that people who run from reality used cannabis, but that's far from being causal.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@simonmaechling I distrust both, because both were paid to give a plausible answer to back established narratives. I trust science because it's few facts are testable.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@avrldotdev I would have another bucket in mem for patterns that slipped. Of course the mech would be harness driven. I would be more concerned that the harness has an open world problem leading to the miss in the first place.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@yashmp2004 Windows and Docker is a horrible experience. Linux or Mac with Podman especially if k8s is your target.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@nathanmarz If it's something a Sr has to correct/nudge then the jr/solo will learn as we did. I didn't need a decade od boiler plate to develop tastes around arch and failures. I think I would have learned the lessons much faster and would have a great notebook.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@CharliLaFerrari If boilerplate was your jam that might be. I am loving how cheap the data driven path is. I can afford to experiment with different solutions with fantastic telemetry.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@Taniyatweets_ 128gb is a good start.

[Systems Pimp 🎩🦯🕶️ @Hunters_laptp]

@theo Dogfooding is required.