Este bootcamp nos costó mucho trabajo prepararlo, pero @haro_ca_ y yo estamos muy orgullosos de cómo quedó.
Pueden pensarlo como un curso de diseño de sistemas centrado en la capa de datos de las aplicaciones, que normalmente son las bases de datos.
💥 ¿Te perdiste el inicio del Bootcamp de Bases de Datos Avanzado? Tenemos algo que contarte.
🔎 ¡Publicamos los primeros 27 minutos en un video no-listado! 🚀 Héctor Patricio, Tech Lead en Autómata, comparte todo lo que puedes esperar del programa.
💬 Comenta si quieres el link 👀
¡Aún puedes unirte! Dejamos abiertas las inscripciones semana más. Solo hemos cursado la sesión de inauguración (que ya puedes ver grabada) y quedan el resto de los 6 meses de sesiones sabatinas de 3 hrs.
¿Listo para sumarte? https://t.co/waB2q29YyT
@leonpalafox Jajaja sinceramente me costó entender lo que escribiste, pero estoy completamente de acuerdo: Jeff Dean ha construido los entornos que hoy son la infraestructura de lo que hacemos. Ignorar su opinión sería tonto.
You know what other tools know better than most instructors? Coursera and YouTube courses from top faculty, *the internet*, books from the library. How many students used those tools instead of formal ed? Very very few. How many will use Claude independently to learn the material? Probably the same amount.
I know it doesn’t sound glamorous, but the primary role of faculty is to get students in the seats and create incentives to actually absorb the information. This is your job. AI can help as a tool, I’ve seen some great harnesses of AI for education, but it will not do this.
I've got an agent in a loop optimizing a renderer with the goal to minimize frame times (and tests to measure). It got times down from 88ms to 2ms and allocations down from ~150K to 500. Sounds good, right? Wrong. This is exactly why agent psychosis is a big fucking problem.
As an experiment, I rewrote the Ghostty core render state in Go, with access to identically laid out data structures as Ghostty and the exact same validation tests. I made a purposely naive renderer (simple, correct, but slow). 88ms per frame with 150,000 allocations (horrendous, lol)!
I then kickstarted a Ralph loop to bring the frame times down. I told it it can't modify input data structures or the public API or tests (they're correct), but it can do anything else it wants. It got to work.
It has worked for about 4 hours. I've spent around $350 on this experiment so far. The results?
88ms => 1.5ms
150K allocs => ~500 allocs
Incredible right? Nope.
My hand-written renderer I ported has frame times (same benchmark) of ~20us (0.020ms) and 0 allocations in the update path.
This is the problem with psychosis and lacking systems understanding. If you don't understand the system, you're going to accept that this is an incredible result. If you understand the system, you'll see better solutions immediately and can do roughly 75x better on throughput.
The people who blindly trust agent output are in the former camp. They're sheeple, overdrinking from a fountain of mediocrity.
Standard disclaimer: I use AI all the time. I like AI. The point I'm making is to not blindly accept results. Think. Analyze. Learn.
@DiPietrantonioJ Por eso hasta los shampoos traen instrucciones. Si le das libertad absoluta te va a hacer la porquería más fea del mundo con todo lo que “leyó”.
Whenever anyone complains about vibeslop I remember all the “production” codebases of billion dollar companies I’ve worked with.
It’s so rare to find a repo as clean as something like Pi.
Share your best oss codebase and I will poke around in there. (:
@Jonatha8808511 Completamente de acuerdo. Siento que los ingenieros experimentados vibecodeando son como músicos experimentados probando la improvisación en el jazz. Sobre todo aquellos que tienen nociones fuertes de producto y cultura amplia.