iiinova — where your digital self keeps evolving

Daron Yondem @daron.me · 06/08/2025

🚀 Apparently TSA missed the “3-oz limit on AI models” in my carry-on. #dadjoke #staywithme Thirty days after joining AWS, OpenAI’s brand-new open-weight models, gpt-oss-120b and gpt-oss-20b, just landed in Amazon Bedrock and SageMaker. Total coincidence…

111

Daron Yondem @daron.me · 28/07/2025

Attaching my race map and official time for anyone who loves the nerdy details 🙂

100

Daron Yondem @daron.me · 28/07/2025

Fast-forward to this morning: same city, same Golden Gate views, but a very different ending. Crossing that finish line wasn’t just about a medal or a stopwatch; it was a reminder.

100

Daron Yondem @daron.me · 22/07/2025

💡 Bigger isn’t always better: why I’m reaching for smaller models in my everyday work When I first got access to O3, its raw power blew me away, I honestly thought I’d never settle for anything less.

100

Daron Yondem @daron.me · 10/07/2025

On Saturday, August 16 I’m trading weekend plans for something way more fun: the DeepSeek Demystified Summit, a one-day, all-access dive into the open-source LLM that’s making waves across cost, performance, and reasoning. Why I’m excited:

100

Daron Yondem @daron.me · 08/07/2025

Want to see it? I’m dropping the playground link + a fun speed-up video (video effect exaggerated for Twitter attention 😉).

100

Daron Yondem @daron.me · 07/07/2025

🤯 What if your Transformer paid attention to triangles instead of lines? Those extra corners buy you token efficiency and a steeper scaling curve.

220

Daron Yondem @daron.me · 14/06/2025

Just wrapped up my first week with the Bevel Health app, and I’m honestly blown away. 📱✨

200

Daron Yondem @daron.me · 23/05/2025

🚀 Counting down to European AI and Cloud Summit next week! I’m excited to head to Düsseldorf for the European AI & Cloud Summit (26-28 May) where I’ll be taking the stage to share some fresh insights on Multi-Agent AI Workflows.

220

Daron Yondem @daron.me · 22/04/2025

Smart and lean: KnowSelf cuts costs with fewer knowledge calls and boosts generalization. Trained on three tasks, it aced unseen ones like Heat and Cool. Self-awareness shines with only 40% training data—perfect for small models.

100

Daron Yondem @daron.me · 22/04/2025

🤯 What if AI agents could sense when they’re out of their depth—pausing to reflect or fetch knowledge before acting? Meet “KnowSelf,” a new paradigm for LLM agents. Thread below 👇

120

Daron Yondem @daron.me · 18/04/2025

👉 Self-critique boosts quality. AAG’s model critiques its own drafts for up to 3 cycles, suggesting edits and rewriting. This lifts answer quality by 7-10%, proving “judge-and-fix” beats bigger prompts. #AIResearch

100

Daron Yondem @daron.me · 18/04/2025

🚀 Large language models excel at facts, but can they craft plans for brand-new tasks? Introducing Analogy-Augmented Generation (AAG)! #AIResearch #LLM

230

Daron Yondem @daron.me · 17/04/2025

💰 Cost frontier: o4-mini hits 93.4% on AIME ’24 at ~⅓ GPT-4o latency. o3 saves dev time—2h/week vs o1, per my CI. #Efficiency

100

Daron Yondem @daron.me · 17/04/2025

🔗 Function-calling 2.0: Responses API keeps reasoning tokens, ensuring reliable tool use. No more shaky JSON regex! #LangChain

100

Daron Yondem @daron.me · 17/04/2025

🛠️ Tool-selection head picks web. run, python_user_visible, or image_gen. Scores 24.9% on Humanity’s Last Exam with tools vs 20.3% without. #AIEngineering

100

Daron Yondem @daron.me · 17/04/2025

⚡️ OpenAI’s o3 & o4-mini are here, moving us from chatbots to autonomous analysts. These models browse the web, run Python, process images, and plan actions. #AI #OpenAI

120

Daron Yondem @daron.me · 15/04/2025

💰 Efficiency alert: GPT-4.1 is 26% cheaper than GPT-4o. GPT-4.1 mini matches/exceeds GPT-4o smarts at half the latency, 83% lower cost. ⚡ #TechNews

110

Daron Yondem @daron.me · 15/04/2025

📚 Context window now hits 1 million tokens (from 128,000), with precision throughout. Huge win for big docs and long chats! #AI #Innovation

100

Daron Yondem @daron.me · 15/04/2025

🤖 Instruction following? Up 10.5% in reliability. GPT-4.1 tracks your requests deep into convos – a must for real-world apps. #MachineLearning #APITechnology

100

Daron Yondem @daron.me · 15/04/2025

💻 GPT-4.1 crushes it in coding – scoring 54.6% on SWE-bench Verified, a 21.4% leap over GPT-4o. Better repo understanding, reliable task completion, and working code! #Coding #OpenAI

100

Daron Yondem @daron.me · 15/04/2025

🚀 OpenAI just launched GPT-4.1, GPT-4.1 mini, and GPT-4.1 nano – a new family of models with stellar performance upgrades. Available now via API only, not in ChatGPT yet! #ArtificialIntelligence #GPT4.1

220

Daron Yondem @daron.me · 14/04/2025

The real game-changer? Agentic-Tx. It reasons about molecules, searches literature, and tests hypotheses, scoring 52.3% better on tough therapeutic reasoning benchmarks. This is AI that *thinks* like a scientist! #TherapeuticDevelopment

100

Daron Yondem @daron.me · 14/04/2025

🧬 Google DeepMind's TxGemma doesn't just predict drug molecules—it EXPLAINS why they might work! This could unlock a major bottleneck in AI-powered drug discovery. #AIinHealthcare #DrugDiscovery

120

Daron Yondem @daron.me · 13/04/2025

After nearly 3 years away from racing due to injuries, I'm finally back on the course — and it feels amazing! 😊 Ran my first 10K since the break and finished 12th out of 300 in my age group. Not a bad comeback! 🏃‍♂️💨

320

Daron Yondem @daron.me · 11/04/2025

Why I Refused to Play the AI Detection Game. I walked away from a publisher over their obsession with "AI detection" thresholds. They wanted me to tweak my original work to dodge some arbitrary % limit. I’d rather focus on quality than dance for a tool.

120

Daron Yondem @daron.me · 10/04/2025

What if neural networks were a Wonderland, and we're all just curious Alices trying to make sense of it? Just finished Scardapane's brilliant 'Alice's Adventures in a Differentiable Wonderland' and I'm still processing how elegantly it demystifies the rabbit hole of neural networks.

220

Daron Yondem @daron.me · 09/04/2025

The same scaling laws that boost LLMs apply here: performance rises consistently as model size grows, just like language models. A striking parallel in AI development! #AIforRecommendation

100

Daron Yondem @daron.me · 09/04/2025

They’re "tokenizing" your viewing history, turning binge sessions into sequences that capture preferences. Unlike LLMs where tokens are equal, Netflix weights a 2-hour movie watch more than a 5-min trailer browse. #MachineLearning

100

Daron Yondem @daron.me · 09/04/2025

Netflix is revolutionizing recommendation systems by applying Large Language Model (LLM) principles at scale. Their foundation model processes hundreds of billions of user interactions—comparable to the token volume in ChatGPT. #AIforRecommendation #FoundationModels

110

Daron Yondem @daron.me · 07/04/2025

Even if 10M tokens worked (doubtful), RAG wins for enterprise. It’s leaner, faster, and nails 99% of knowledge tasks without the overhead. #RAG #GenerativeAI

100

Daron Yondem @daron.me · 07/04/2025

LLaMA 4’s 10M token context window sounds wild—bye-bye vector databases and RAG? Not quite. The hype is big, but the story’s more complicated. Let’s break it down. 🧵 #LLMs #AI

220

Daron Yondem @daron.me · 05/04/2025

The last time I took the Scrum Master I exam was back in 2012 😊 Yesterday, I figured I’d give it a shot—and I passed Scrum Master II ! Turns out, I really didn’t need to wait this long. The questions were solid. I’d say anyone with 3–5 years of Scrum experience should be able to pass pretty easily.

020

Daron Yondem @daron.me · 04/04/2025

Not every AI problem needs a multi-agent system. Most don’t. Chasing complexity too early wastes time & burns trust. #CTO #GenAI #MultiAgentSystems

220

Daron Yondem @daron.me · 03/04/2025

🤯 72% of professors fear AI lets students skip thinking. But Claude for Education’s new Learning Mode flips the script—designed to spark learning, not shortcut it. #AIinEducation #EdTech

140

Daron Yondem @daron.me · 02/04/2025

🚀 BREAKING: A paralyzed woman speaks again after 18 years—using only her thoughts! 🧠✨ Her own voice, powered by a brain implant & AI, communicates in real-time. Not sci-fi, but science. 👇

210

Daron Yondem @daron.me · 01/04/2025

🚨 Big news in mental health: An AI therapist achieved human-level therapeutic alliance in half the time, reducing depression by 51%, anxiety by 31%, and eating disorder concerns by 19% in a groundbreaking clinical trial. 🧵

110

Daron Yondem @daron.me · 31/03/2025

🧠 Can an 8B open-source model outperform GPT-4 in collaborative tasks? Turns out—it can. But only with the right training recipe. 👇

230

Daron Yondem @daron.me · 28/03/2025

🌍 The ecosystem is growing fast with tools like Mintlify’s mcpt and Claude Desktop. Implications: - Tools as scenario-driven APIs - From prompt hacking to tool orchestration - New hosting/auth models

100

Daron Yondem @daron.me · 28/03/2025

🔍 Real-world example: Your IDE pulls logs from a Browsertools server, queries Postgres via MCP, and generates an image through Replicate—no app switching or glue code needed. That’s MCP in action.

100

Daron Yondem @daron.me · 28/03/2025

💡 Inspired by LSP, MCP enables: - Autonomous tool chaining by agents - Human-in-the-loop control - Seamless integration with dev tools and SaaS It’s like an OS for agent workflows, replacing brittle integrations.

100

Daron Yondem @daron.me · 28/03/2025

🚀 MCP: The Protocol Poised to Revolutionize AI Tool Integration Imagine a world where AI agents autonomously select and sequence tools without manual coding. That’s the promise of the Model Context Protocol (MCP)—a new foundation for AI workflows.

110

Daron Yondem @daron.me · 26/03/2025

For those interested, attached is how I create the first image in this thread. Does it look like me?

010

Daron Yondem @daron.me · 26/03/2025

Was deep into prompt engineering this morning when I stumbled upon OpenAI's latest: the 4o Image Generation model. It's incredible! Here’s why: #AIImageGeneration #OpenAI #GPT4o #PromptEngineering #AITechnology

111

Daron Yondem @daron.me · 25/03/2025

🚀 Anthropic just dropped a major upgrade for Claude: the "think" tool! This lets Claude pause, reflect, and tackle complex, multi-step tasks with ease. Developers integrating LLMs into apps, this is for you—here’s why it’s a big deal. #AI #LLM #DeveloperTips

120

Daron Yondem @daron.me · 24/03/2025

🔍 In high-stakes scenarios, wrong AI answers can be costly. New research from Johns Hopkins shows "test-time scaling" boosts accuracy and helps models know when to stay silent.

110

Daron Yondem @daron.me · 21/03/2025

Check the sample audio below—hear the expressiveness! It’s like giving your AI a personality upgrade. 🎭 So cool!

100

Daron Yondem @daron.me · 20/03/2025

Another day, another agentic framework! NVIDIA's AgentIQ treats agents, tools, and workflows as simple function calls for true composability. Build once, reuse everywhere. 🔄 #AIEngineering #LLMOps

120

Daron Yondem @daron.me · 19/03/2025

Welcome Eclipse LMOS - an open-source, cloud-native platform laying the groundwork for an "Internet of Agents." 🌐 Think of it as connecting AI agents across orgs, just like the Internet linked computers and IoT tied devices. #AIAgents #OpenSource

100

Daron Yondem @daron.me · 18/03/2025

Baidu just dropped ERNIE 4.5 & ERNIE X1, pushing AI to new heights! 🚀 ERNIE 4.5 rivals GPT-4.5 in benchmarks but at a fraction of the cost. ERNIE X1 excels in logical reasoning & problem-solving. Big news for AI devs & researchers! #AI #MachineLearning #NLP

100