rafaelero

The current consensus is that the effects of social media on mental health are in between minimally detrimental to non-existent.

Expensive and Google doesn't even have enough resources to decently deploy a model like that. Creating 10 images in parallel gives me RESOURCE_EXHAUSTED error, which is a painfully common error when using Google AI products.

I think it's pretty safe to assume they are not losing money on inference.

Pretty much every person in every culture cares deeply about their children.

I would substitute "deeply" for "superficially". Like, if my parents found some way to prohibit porn when I was an adolescent, I wouldn't say they cared deeply about me. I would say they were misguided and authoritary. The "care deeply" idea you are putting forward is just trying to distil whatever societal norm currently is into the youngs.

Companies who could see it clearly and ignored the "AI is a bubble duh" crowd will ultimately get benefited by the GPUs they already acquired. The companies who acted cautiously will get burned.

Location: Portugal

Remote: Yes

Willing to relocate: Yes

Technologies: Python, Deep Learning (PyTorch), NLP, Graph Knowledge Systems, Vector Search (Cosine Similarity), PostgreSQL, React/Svelte.

Résumé/CV: https://www.linkedin.com/in/rodrigo-heck-7280a218a/

Email: rodrigo.heck29@gmail.com

Hi HN, I’m Rodrigo. I’m a software engineer working at the intersection of applied ML and scalable systems. Lately, I’ve been obsessed with solving the "context window" problem through graph-based knowledge representations. My recent work focuses on treating graphs not just as data stores, but as a method for information compression and long-term memory permanence in AI agents. I develop pipelines that integrate LLMs with structured memory to allow for more efficient knowledge generalization and recall.

Key areas of expertise:

1. AI Memory Architectures: Building persistent, graph-oriented systems for reasoning and long-term retrieval.

2. ML Systems Engineering: Developing TTS systems, semantic search tools, and RAG pipelines that go beyond simple vector lookups.

3. Full-Stack Foundations: Bridging the gap between a PyTorch model and a production-ready Svelte/React interface, backed by robust Linux/Postgres infrastructure.

I’m looking for a role where I can contribute to the "Next Step" of LLM integration—moving past simple chat interfaces toward systems with true persistent memory and structured reasoning.

I like you.

Last time I checked there's no scientific consensus if social media causes harm at all. The best studies found null or very small effects. So yeah, I am skeptical it is harmful.

I have no idea where this idea that Internet is toxic to children is coming from. Is that some type of moral panic? Weren't most of you guys children/adolescents during the 2000's?

The equivalence of emotions to reward functions seem pretty obvious to me. Emotions are what compel us to act in the environment.

Idk, we seem to be the at the cusp of autonomous driving. Transportation is like ~8% of world's GDP. Payroll is, what, 30% of that? It seems like we can already have the return on all AI investment by just conquering this one application.

Yeah, I suspect the reason the author didn't find a relationship between IQ and happiness / life satisfaction is probably because those studies were overcontrolling for intermediate variables. If money makes us happier and people with high IQ make more money, you will underestimate the relationship if you control for income.

Location: Portugal

Remote: Yes

Willing to relocate: Sure

Technologies: Deep Learning (PyTorch), ReactJS, Svelte, Natural Language Processing (NLP), Python (Flask, OpenAI API), Graph Knowledge Systems, Vector Search (Cosine Similarity), PostgreSQL, Linux System Administration

Résumé/CV: https://www.toptal.com/resume/rodrigo-heck

Email: rodrigo.heck@toptal.com

Lately, I’ve been exploring graph-based knowledge representations as a method for information compression and long-term memory permanence in AI systems — developing pipelines that integrate LLMs with structured memory to retain and generalize knowledge efficiently.

My background bridges applied machine learning and software engineering, from building text-to-speech systems and semantic retrieval tools to experimenting with persistent, graph-oriented architectures for reasoning and recall. I’m looking for opportunities at the intersection of AI systems engineering, knowledge representation, and scalable memory architectures.

Contributing to human welfare through technological advances and productivity gains is a much better place to deposit my hopes and emotions, imo.

It's honestly not that deep. If AI increases productivity, we should accept it. If it doesn't, then the hype will eventually fade out. In any case, having attachment to the craft is a bit cringe. Technological progress trumps any emotional attachment.

Looks like you are correct.

The problem with this approach to text generation is that it's still not flexible enough. If during inference the model changes its mind and wants to output something considerably different it can't because there are too many tokens already in place.

Things are already going severely wrong in 1% of the cases. At this point not getting a second opinion from an LLM is irresponsible, imo.

I did create benchmarks and simulated impact following different scenarios (including the one I was advocating for). Unfortunately that was received in deaf ears. Even worse, actually; they thought I was being too pushy by adding the scenario I was proposing to the analysis. At this point I knew I had to leave.

Even with all these tools available, big companies would still be unable to compete in speed simply because in 99% of cases they don't have the required culture set in place.

This is a good strategy, imo. I was following it for almost a year and I was having a blast working in a startup. Then a new manager came along and started to dictate how things should be done without much input from the technical team. I kept fighting for what I thought was good for the product instead of aligning with him. In the end it was just too stressful (the manager was not only an idiot but also rude). I resigned, but I wouldn't have done any other way. I simply can't be made to do dumb things from uncurious people.

Food can also indirectly enhance opioid signaling. What's your point?

Tbh, it's not clear that if we all stopped drinking the world would be better.

This new Gemini Flash 2.5 is cutting the response in the middle. Did anyone experience that?

To be fair, even if you understand it that's doesn't mean it will work well.

It compares AI-adopting companies with non-adopting firms, so that should account for the overall economic scenario.

Why do you have to imply there's some conspiracy? Haven't you noticed productivity gains by incorporating AI to your workflow? If you have, then there's your answer: we point to that possibility because we are seeing everyday how much more we are accomplishing by using AI. If demand doesn't increase as much, then there will be a decrease in employment.

Nothing in the rate of improvement of a steam engine suggest it would be able to drive a car or do the job of an attorney.

I feel like you are trapped in the first assessment of this problem. Yes, we are not there yet, but have you thought about the rate of improvement? Is that rate of improvement reliable? Fast? That's what matters, not where we are today.

god they keep raising prices