heidrian.eth | .lens | .rsk

@paubgood Hola Paula. Paso tu contacto al equipo, me interesa.

¡Hola a todos/as! Soy economista y data scientist con foco en métodos cuantitativos, evaluación de políticas. Uso R/Python e integro otras herramientas para hacer análisis rigurosos y también proyectos creativos (moda, Taylor Swift, etc.). Busco roles donde los datos se conviertan en decisiones: data science/analytics, evaluación de impacto, research y data storytelling. Pueden ver mi portfolio y CV en mi blog: paulapereda.com Estoy buscando un nuevo puesto y agradecería apoyo. Si alguien sabe de alguna oportunidad o simplemente quiere que nos pongamos al día, puede enviarme un mensaje.

@mrlesk @karpathy It's like micro-managing multiple people, at some point your brain becomes the bottleneck

In regard to cc/codex running in parallel: their limits are not too different from human ones. Think in terms of context and conflicts. If you have some tasks that are not touching the same files, don’t need to share the same context then they are good candidates for parallelism.

Continuing the journey of optimal LLM-assisted coding experience. In particular, I find that instead of narrowing in on a perfect one thing my usage is increasingly diversifying across a few workflows that I "stitch up" the pros/cons of: Personally the bread & butter (~75%?) of my LLM assistance continues to be just (Cursor) tab complete. This is because I find that writing concrete chunks of code/comments myself and in the right part of the code is a high bandwidth way of communicating "task specification" to the LLM, i.e. it's primarily about task specification bits - it takes too many bits and too much latency to communicate what I want in text, and it's faster to just demonstrate it in the code and in the right place. Sometimes the tab complete model is annoying so I toggle it on/off a lot. Next layer up is highlighting a concrete chunk of code and asking for some kind of a modification. Next layer up is Claude Code / Codex / etc, running on the side of Cursor, which I go to for larger chunks of functionality that are also fairly easy to specify in a prompt. These are super helpful, but still mixed overall and slightly frustrating at times. I don't run in YOLO mode because they can go off-track and do dumb things you didn't want/need and I ESC fairly often. I also haven't learned to be productive using more than one instance in parallel - one already feels hard enough. I haven't figured out a good way to keep CLAUDE[.]md good or up to date. I often have to do a pass of "cleanups" for coding style, or matters of code taste. E.g. they are too defensive and often over-use try/catch statements, they often over-complicate abstractions, they overbloat code (e.g. a nested if-the-else constructs when a list comprehension or a one-liner if-then-else would work), or they duplicate code chunks instead of creating a nice helper function, things like that... they basically don't have a sense of taste. They are indispensable in cases where I inch into a more vibe-coding territory where I'm less familiar (e.g. writing some rust recently, or sql commands, or anything else I've done less of before). I also tried CC to teach me things alongside the code it was writing but that didn't work at all - it really wants to just write code a lot more than it wants to explain anything along the way. I tried to get CC to do hyperparameter tuning, which was highly amusing. They are also super helpful in all kinds of lower-stakes one-off custom visualization or utilities or debugging code that I would never write otherwise because it would have taken way too long. E.g. CC can hammer out 1,000 lines of one-off extensive visualization/code just to identify a specific bug, which gets all deleted right after we find it. It's the code post-scarcity era - you can just create and then delete thousands of lines of super custom, super ephemeral code now, it's ok, it's not this precious costly thing anymore. Final layer of defense is GPT5 Pro, which I go to for the hardest things. E.g. it has happened to me a few times now that I / Cursor / CC are all stuck on a bug for 10 minutes, but when I copy paste the whole thing to 5 Pro, it goes off for 10 minutes but then actually finds a really subtle bug. It is very strong. It can dig up all kinds of esoteric docs and papers and such. I've also used it for other meatier tasks, e.g. suggestions on how to clean up abstractions (mixed results, sometimes good ideas but not all), or an entire literature review around how people do this or that and it comes back with good relevant resources / pointers. Anyway, coding feels completely blown open with possibility across a number of "kinds" of coding and then a number of tools with their pros/cons. It's hard to avoid the feeling of anxiety around not being at the frontier of what is collectively possible, hence random sunday shower of thoughts and a good amount of curiosity about what others are finding.

@bourbonni si el rainbow dice que hay que vender, se vende. Eso no quiere decir quedarse en dolares, hay formas mas tradicionales de protegerse de la inflación (tips, real estate, etc.) hasta que sea momento de volver a comprar BTC.

Si tenés bitcoin, hay algún precio al cual venderías todo?. No hablo de “una parte” sino de literalmente ya no quedarte con ningún sat?

Me parece interesante esto. En mí experiencia personal el cuello de botella termina siendo la capacidad de mí mente de administrar con éxito N agentes (o sea dar órdenes constantes sobre N temas) en paralelo sin llegar quemado al final del día. Como si micromanageara N empleados.

@danielgothits @RealMetaPrince @inurinternet yeah, sure hope it was destroyed

@RealMetaPrince @inurinternet @danielgothits you may trust "the seal" maybe, but at some point a computer you don't own generated the seed, some machine printed the label and another machine or a person pasted/closed the seal on top of the label

@AdrianHMazza @inurinternet @danielgothits Same thing I've been wondering while reading this, the person who sold or gifted you a bitcoin wallet would obviously have access to the wallet himself. Or am I oblivious to how this works?

@inurinternet @danielgothits May I ask what was the opsec to ensure neither you or your employees knew the sealed seeds. I wouldn't hold value in a preprinted seed whatsoever.

@danielgothits wow! i gave a few employees 1btc casascius coins back then. i often wonder who still has theirs

Tenemos una nueva búsqueda abierta en @Alphacast_ , para uno de los roles más estratégicos de la operación: Lider de Calidad de Datos, quien garantiza que los datos estén siempre correctos, completos y a tiempo. Gente del mundo de los datos, de la economía, de los datos y la economía, pueden ver más acá!! 👇👇👇 linkedin.com/feed/update/ur… Favor RT!!

@ViktorBunin Use the + alias in emails (if using gmail based, maybe it works on other mail servers too) to add a short random salt to email addresses. That way they won't even know which account to try to hack into.

I keep getting emails and texts that folks are trying to reset my passwords on various sites. I have strong security but it's still a bit stressful. Anyone have advice or thoughts on how to deal with this or mitigate it?

@dariovende Felicitaciones!

@AcheronTychus @bate5a55 @aidan_mclau Not saying anything about LLMs, just stating that your "numbers for symbols swap" is no good test to separate mimicking from proto-reasoning

an efficient market would see apple's stock down 10% after publishing this

@bate5a55 @aidan_mclau You should try doing the same to a 5th grader. See if replacing numbers with symbols doesn't degrade their reasoning

@aidan_mclau Did you know the paper shows LLMs fail dramatically when numbers are replaced with symbols in math problems? It suggests they're pattern-matching, not actually reasoning.

@BtcAndres @0xNanocol @gguarapo Se puede usar el AaveUSDC como alternativa, que tampoco tiene freeze. O quedarse en DAI sin migrar.

@0xNanocol @AdrianHMazza @gguarapo Hay más

De cómo mi amigo en Colombia perdió más de un tercio DE MILLÓN DE DÓLARES en un minuto, con sus cRiPtO congeladas en Trust Wallet. Imagina este escenario. Abres tu wallet y no puedes mover todo tu dinero. Aunque, por casualidad, un mes antes me llamaste con este punto de vista: "Andrés, si aproximadamente el 1% de los Tether se congelan, NO CREO que me suceda a mí. Además, pon en una balanza la facilidad de entrar en dólares en estos países latinos, para eso Tether es una gran herramienta, y ahora compara con la probabilidad de que algo 'extraño' suceda con mis monedas". Bueno, nada te pasa, HASTA QUE TE PASA. Este escenario no es ficticio. HA SUCEDIDO. Me llama mi amigo y me dice que tiene $348.910 USDt congelados en Trust wallet. Y encima me recuerda mis palabras: "Andrés había dicho que no tus llaves privadas, no tus monedas". OJO: Pero también te había dicho que esto solo aplica para bitcoin, no para cRiPtO (en muchísimos videos), sin entrar a hablar de wallets de código cerrado que no deberías usar como Trust wallet o Exodus wallet. Mi amigo intenta enviar 100 USDt, 1.000 USDt y luego 500 USDt. Nada. Nada se mueve. Hablando de wallets de código cerrado como Trust o Exodus, por primera vez vas a ver cómo tu wallet aparece con una hermosa alerta diciendo que tus fondos vienen de alguna actividad ilícita (en bitcoin ya sabemos que no aplica porque todo lo que hacemos es ilegal en algún lugar del planeta, por lo que no hay reguladores, pues en algún momento preguntaríamos QUIÉN REGULARÁ A LOS REGULADORES?). Mira la foto: Recuerda. Este bloqueo no lo hizo Trust Wallet. Lo hizo Tether en una de las blockchains que corre. Puedes usar lo que quieras. Pero recuerda tener un poco de cuidado allá afuera. O no. Son tus cRiPtO.

@Ariperelman 👏🏻👏🏻👏🏻👏🏻👏🏻

En un par de meses vamos a estar proveyendo seguridad a la blockchain de Bitcoin convirtiendo gas metano cancerígeno que las compañías petroleras ventean a la atmósfera, en energía limpia, segura y sustentable ⚡️♻️ Cómo lo hacemos?

@BtcAndres @gguarapo En gran parte USDC pero por la naturaleza de la componibilidad no hay posibilidad desde el sc de USDC de bloquear a un tenedor específico de DAI, cosa que hizo Tether en el caso que mencionas. Deberían bloquear a todo DAI (al PSM), y eso es comercialmente impracticable.

@AdrianHMazza @gguarapo Cuál es el activo subyacente?

@BtcAndres @gguarapo No, DAI no se puede blacklistear por address. No tiene ese método el smart contract.

@gguarapo twitter.com/btcandres/stat…

@CarlosMissirian Ah, si, debe ser verso

@AdrianHMazza Obvio pero a) el rol era DE NEGOCIO/PRODUCTO y b) se referia a gen ai Te diría que si hubiera dicho machine learning no me jode pero esta muy raro porque honestamente era todo academia y no habia muchas aplicaciones de negocio.

Para que vean que LinkedIn es un lugar sin lógica: