Isabell Geller

The line between traditional markets and crypto keeps getting thinner. Tokenized assets and onchain perps are making finance more accessible than ever. Canborsa is one of the projects I'm watching in this space

@realBigBrainAI We understand how to train it. We're still learning how it truly thinks

Inside the "brain" of AI: billions of parameters, one prediction, and a path no one can fully map. Source: explainingtheuniverse

@PythonPr Funny meme, but most real-world systems end up using all three at some point

C++ vs Java vs Python 🤣🤣🤣

@voidshivendra Nice work. The learning never stops in tech, and that's what makes it exciting

@PythonDvz RAG solved access to knowledge. Agentic AI is solving execution. That's where the next wave of value comes from

Different Types of LLM Algorithms and Architectures: A Practical Guide for AI Professionals 🚀 Large Language Models (LLMs) are transforming how we interact with technology. From intelligent chatbots and code assistants to enterprise search and AI agents, LLMs are becoming the foundation of modern Artificial Intelligence. But not all LLMs are built the same. Understanding the major LLM architectures helps Data Scientists, AI Engineers, and Technology Leaders choose the right model for the right problem. Future Trends The future of LLMs is moving toward: Multimodal AI (text, image, audio, video) Agentic AI systems Smaller and more efficient models Enterprise-grade AI applications Personalized AI assistants Key Takeaways ✅ Transformers remain the foundation of modern LLMs. ✅ Different architectures serve different business needs. ✅ RAG improves factual accuracy through external knowledge. ✅ Agentic AI is pushing beyond simple text generation toward autonomous decision-making. As AI continues to evolve, understanding these architectures is becoming a critical skill for anyone working in Data Science, Machine Learning, or Artificial Intelligence. 💬 Which LLM architecture do you think will have the biggest impact on enterprise AI over the next five years? #ArtificialIntelligence #GenerativeAI #LLM #LargeLanguageModels #RAG #AgenticAI #MachineLearning #DataScience #AIEngineering

@DeeAnnaNagel The intersection of AI and mental health is one of the most important conversations happening right now

AI and mental health - here are some great resources for therapists. apa.org/topics/artific…

@DrJimFan Physical AI gets much more interesting when agents can learn directly from reality instead of just datasets. Huge step forward

Today, we enable AutoResearch in the physical world for the first time! Introducing ENPIRE: we give 8 Codex agents a fleet of robots, an allocation of GPUs, and generous token budget. We set them free with a simple goal: solve the task as quickly as possible, keep the robots busy but stay safe, don't waste precious compute. Make no mistake. Then humans step aside and our watch begins. The robot fleet starts to come alive: they learn to look for visual clues, reset the scene, practice novel skills, tinker with control stack, read papers online, debate, reflect, get stuck, and try again directly on the hardware. All we did is to give Codex an API to the world of atoms, and the rest is emergence. ENPIRE is able to solve high-precision tasks like tying zip-ties, organizing fine pins, and installing GPUs all by itself. We also discovered a new type of "physical scaling": 8 robots exploring in parallel improves significantly faster than fewer ones. A part of our NVIDIA GEAR lab now self-improves tirelessly over night. We just read the reports in the morning. /goal: we all take a holiday and Jensen wouldn't even notice ;) We will be open-sourcing everything, so you can host your self-running robot lab at home too! Deep dive in the thread:

@WenergyXR Digital Twins could become one of the most practical applications of AI in industry over the next decade

Sensors → Connectivity → Data Platforms → Visualization → Analytics → AI Individually, they're technologies. Together, they create Digital Twins that deliver real-world value through visibility, prediction, and optimization. #DigitalTwin #AI #IoT #WenergyXR

Interesting to watch how AI is moving from experimentation to real business applications

@priya_Thakur786 C++ for me. Still remember that first "Hello World" :)

What Language Did You Use for Your First "Hello World"?

@komal_uk01 ChatGPT for reasoning. Claude for large codebases

As a developer, which AI is the best for coding?

@RBFMorph 25% weight reduction while meeting all structural requirements is impressive work. Congrats!

🎓 Congratulations to Gabriele Bortolotto on earning his Engineering degree from the University of Rome Tor Vergata with the successful defense of his thesis: "Finite Element-Based Thermo-Structural Optimization of an Internal Combustion Engine Intake Manifold" The research leveraged RBF Morph integrated within ANSYS Mechanical to perform advanced mesh morphing and parametric optimization of a high-performance automotive intake manifold. By modifying the finite element model directly without remeshing, multiple design alternatives were evaluated efficiently, leading to a final design that achieved a 25% weight reduction while satisfying all structural requirements. We are proud to see RBF Morph supporting innovative academic research and helping the next generation of engineers develop the technologies of tomorrow. Congratulations, Gabriele, and best wishes for your future career! You can read the thesis here: rbf-morph.com/leveraging-rbf… #RBFMorph #Engineering #CAE #FEA #ANSYS #EngineeringOptimization #AutomotiveEngineering

@FredFooe The scary part isn't AI generation. It's AI manipulation at scale

1. Core Algorithm Developers: These entities occupy the deep technical tier. They focus on basic computer vision research, optimization of transformer blocks, and the creation of low-latency model architectures. They rarely engage in the direct deployment of campaigns, instead distributing their work through open-source repositories or private commercial licensing. 2. Pipeline Integrators: These operators bridge the gap between technical code and functional deployment. They build the custom software suites that connect AI models directly to streaming interfaces (such as OBS Studio or vMix). They configure the automated masking rules, light field estimators, and asset libraries necessary for real-time manipulation. 3. Front-Facing Personas (The Presenters): These are the public faces of the distribution chain. They include influencers, commentators, or automated virtual avatars who host the live streams. In many operational profiles, the presenter may be entirely unaware that the back-end engineering team is modifying their background or real-time presentation elements to fit specific target narratives. 4. Amplification Networks (Syndicated Distribution): This tier consists of highly coordinated bot swarms, secondary content clipping channels, and automated amplification profiles. Their primary function is to ingest the manipulated live feed, extract key high-impact segments, and cross-post them rapidly across multiple social networks to lock in public perception before verification can occur. #ad automated Ingestion and Modification Pipelines In industrial commercial applications—such as dynamic regional advertising or rapid localized marketing—live streams are processed via Automated Ingestion Pipelines. These systems utilize cloud-native infrastructure to dynamically rewrite video elements based on the viewer's demographic profile or geographic location. ┌──> Target Cohort A ──> [AI Modification Engine A] ──> [Stream A] [Raw Master Stream] ──┼──> Target Cohort B ──> [AI Modification Engine B] ──> [Stream B] └──> Target Cohort C ──> [AI Modification Engine C] ──> [Stream C] A raw master stream is transmitted from a studio to a central cloud architecture (e.g., AWS, Google Cloud, or Microsoft Azure). As the stream is replicated across various distribution nodes, automated computer vision scripts evaluate the video content. If the script identifies a designated "substitution zone" (such as a generic beverage container on a table or a poster on a wall), it triggers localized inference models. The system instantly swaps the asset out—replacing it with a localized brand, alternative textual copy, or specific financial indicators—tailoring the live reality to distinct viewing audiences simultaneously. Section 5: Tracking Frameworks, Analytical Hashtags, and Forensic Countermeasures As live video manipulation technologies advance, digital forensic experts, open-source intelligence (OSINT) analysts, and security researchers have established specialized frameworks to categorize, track, and expose tampered visual media. #Taxonomy of Digital Manipulation Identifiers (Hashtags and Meta-Labels) In the digital research ecosystem, specific semantic labels and hashtags are utilized to aggregate findings, index research papers, and flag suspicious media streams. These labels serve as critical reference points for identifying manipulation methods: * #Deepfake / #SyntheticMedia: The overarching architectural categories used to classify any video or audio stream where human likeness, environmental context, or voice data has been generated or heavily altered using deep learning models. * #DiminishedReality / #ObjectRemoval: Technical designations focused specifically on the erasure of physical matter from video feeds.

@OutlierBlackBox The best opportunities are often gone by the time retail gets access

The ONLY problem with capitalism and markets, is that there isn't much more democratisation of investments. VCs, banks and Private Equity monopolising access to the best companies until very late stages and engineering extra high valuations before 'allowing' access.

@williams_qayode Fiscal priorities are macroeconomics too

They said it's macroeconomics that's being focused on

@themanebroo Understanding macroeconomics doesn't mean supporting every politician

It’s just shameful. Same person seeing positives in macroeconomics, you guys will be thinking about nation state economics in foreign investor language. Please get your PVC and sing on your mandate. You can’t even fool yourself, talk more of others.

@Anas_founder Europe has Mistral. Africa has opportunity.

USA has Claude USA has ChatGPT USA has Gemini USA has Grok China has Qwen China has DeepSeek China has Kimi China has MiniMax Europe has? Africa has?

@PythonDvz Data nerd pretending to understand markets

Who are you? 🤣

RWA adoption feels different this cycle. It's not just crypto building for crypto anymore

@improvemypage @X AI is moving fast. The people building today will have a huge advantage tomorrow

Hey @X algo, I’m quite new here and I’d love to connect with people interested in: - SaaS - Web design - AI - Automation - Software development - Web development - Business If you’re into that or anything related, drop your project below 👇 and let’s connect