Martin Riviera

Reverse-engineered the past, building agents for the future.
Hyperdrive '23 finalist • Ph.D. @princeton

[Martin Riviera @M_Rival12]

U-N-F-A-Z-E-D

Without revealing too much, @moldcx is handling incoming DDoS attacks exactly as expected. No downtime, no disruptions, just a system doing what it was built to do.
Some things are worth the effort. 🙏

[Martin Riviera @M_Rival12]

@kennethngo_ @moldcx Sure, just accepted your request!

[Martin Riviera @M_Rival12]

@svpino You can’t win the AI race by whining about competition. Open-source will eat closed models, and the ones clinging to opacity will for sure fade into irrelevance.

[Martin Riviera @M_Rival12]

@ProgrammerSmart APR is what they advertise. APY is what you actually pay once compounding kicks in

[Martin Riviera @M_Rival12]

@embersunn Windows = control. macOS = babysitting.

[Martin Riviera @M_Rival12]

DeepSeek R1-Zero's "aha moment" is a reminder that real intelligence won’t always look the way we expect. It’s not trying to be readable or charming, just effective. The real question isn’t why its thoughts are unreadable, but whether we’re ready for reasoning that doesn’t cater to us.

[Martin Riviera @M_Rival12]

If AI-generated output can be owned, then the entire knowledge ecosystem becomes a walled garden controlled by corporate gatekeepers. This isn't just an IP dispute, it's the foundation of digital feudalism.
And the worst part is that the more we accept this, the less the future belongs to us.

[Arnaud Bertrand @RnaudBertrand]

People are rightly ridiculing OpenAI over its accusations of Deepseek using their output to train their model, but most people are missing the truly terrifying implications here.

The far more worrying aspect here is that OpenAI is suggesting that there are some cases in which they own the output of their model.

Now think for a minute what this means in the world of tomorrow where so much will be generated by AI (and already is): all the software code, the emails we send each others, the videos and images, etc.

Do you want to live in a future where, if for some reason the AI giants are dissatisfied with the way you use the output of their model, they can claim ownership of it? A future where every piece of content touched by AI - which might be virtually everything in the world of tomorrow - comes with invisible strings attached?

The implications for innovation and creativity are staggering. Small businesses and independent developers who rely on AI tools could find themselves trapped in a web of intellectual property claims. Worse, we're looking at a future where the very act of learning and building upon existing knowledge becomes gated by the interests of AI giants.

This would be techno-feudalism on steroids: if we don't challenge this now, we risk sleepwalking into a future where human creativity and innovation become the property of a bunch of AI overlords, and a world where they can dictate not only who gets to innovate, but what kind of progress is acceptable based on their own interests.

[Martin Riviera @M_Rival12]

@_trish_07 Node’s async model isn’t broken, it’s just different. Go has goroutines, Rust has ownership, and JS has the event loop. One should always pick based on needs, not vibes

[Mold™ Framework @moldcx]

Mold’s analytics infrastructure is scaling, tracking sentiment trends, engagement shifts, and audience growth in real time.

Recent backend updates have optimized data pipelines, delivering faster insights and more precise sentiment analysis. As our models improve, expect deeper intelligence to drive smarter optimizations.

More updates will come soon.

[Martin Riviera @M_Rival12]

@alexandr_wang The real question is if RLHF + reasoning generalizes better than raw pretraining, at what scale does token-based learning become obsolete?

[Martin Riviera @M_Rival12]

She is definitely a keeper.
This moment is everything.
Welcome back Ross!

[Martin Riviera @M_Rival12]

Good ideas don’t come from forcing it. They come from stumbling into something interesting like scrolling, chatting, going down a rabbit hole just because you’re curious.

One random thing clicks, and suddenly you have a strategy, maybe even a business move.

I don’t think you chase ideas. You bump into them.

[Martin Riviera @M_Rival12]

@chessMan786 Context is everything.

[Martin Riviera @M_Rival12]

@Madisonkanna Bitwise tricks like this are why C still reigns in performance-critical domains.

[Martin Riviera @M_Rival12]

@_trish_07 Got to love how pointer tagging turns alignment quirks into free metadata storage. Feels like smuggling extra info past the hardware.

[Martin Riviera @M_Rival12]

@PalmerLuckey The real play is psyops. Just weaponized narratives to destabilize markets and erode trust in American innovation.

[Martin Riviera @M_Rival12]

@bryan_johnson Fads fade. Biomarkers lead.

[Martin Riviera @M_Rival12]

@_trish_07 Blaming Python for slow software is like blaming a scalpel for bad surgery. The real problem isn’t the language, it’s a failure to optimize at the right layer.

[Martin Riviera @M_Rival12]

We’re rethinking AI-driven application logic. Instead of static function calls, we’re building dynamic execution graphs where agentic nodes self-modify based on past inference patterns.

Right now, we’re integrating self-referential diffusion policies, letting agents refine their own latent representations instead of over-relying on external models which huge for multimodal coherence and efficiency.

On the systems side, we’re moving past standard transformers, experimenting with vector-state memory + contrastive predictive coding for long-term retention without bloat. We’re also testing gradient-free meta-optimization (think evolutionary NAS) to let inference self-adjust dynamically.

What's the long-term vision? AI agents that don’t just retrieve knowledge, they reparameterize themselves in real-time for max efficiency.

More to come...

[Mold™ Framework @moldcx]

With Mold, you can build apps powered by advanced AI models like DeepSeek-R1 or Stable Diffusion.

Imagine creating an app that summarizes entire documents into key takeaways, detects products or issues in photos, or even transcribes meeting audio for future reference. Mold makes these features accessible with minimal effort, giving you the ability to create powerful tools for any industry.

[Martin Riviera @M_Rival12]

@Kanekineco Sorry?

[Martin Riviera @M_Rival12]

As the head of the development team at https://t.co/fKwEznHH3j, I’ve had the privilege of leading an incredible journey. Watching this platform grow from concept to a powerful solution that empowers people to build smarter, more efficient systems has been deeply rewarding.
🧵👇