Crunch (Quant/acc)

@GPUMaxxing @herelle_jean username opinion correlation: 100%

@crunchDAO @herelle_jean Mining pools weren’t an era, they were the training wheels. the real evolution is when the pool gets its own gpu and starts mining you.

Every major network undergoes an evolution, and we are officially in Bittensor's Mining Pool Era. An early-stage opportunity akin to the birth of BTC mining pools a decade ago. @herelle_jean will be presenting a masterpiece thesis that literally belongs in the Louvre Palace.

@her6616293 @ReggyStev @numinous_ai @ai_detection @SynthdataCo @mantis Happy to be a green flag enabler for great projects pushing boundaries on Bittensor! 🟩

@ReggyStev @crunchDAO @numinous_ai @ai_detection @SynthdataCo @mantis Great to see some Crunch Numbers Good progress 👏

The list of the agentic intelligence on Bittensor based on 3 parameters, # miners (>50) over 2 weeks, % of burn, good overall score by @crunchDAO: @numinous_ai (0%) (excellent) @ai_detection (0%) (excellent) @SynthdataCo (52%) (excellent) @MANTIS (38%) (excellent) CliqueAI (4%) (good) @Investing88ai (16%) (good) acceptable burn ~50%, acceptable @crunchDAO score - at least "moderate". These are formal parameters but the fact that @crunchDAO mines both @numinous_ai and @SynthdataCo are valid green flags.

Thanks for this, Graham. Indeed, most large enterprises aren't short on data, they're short on high-stakes problem intelligence they can trust to act on. Crunch closes that gap by routing defined problems to a world class global network that competes on measurable performance, not billable hours.

€50 Million Written Off: Ireland’s Public IT Failure Problem When a rail control system consumes tens of millions and still does not deliver the promised capability, the scandal is not software. It is governance. Irish Rail reportedly writing down €50 million on a train traffic management system linked to Indra Group is not just another delayed public IT project. It is a serious indictment of public procurement, vendor oversight, and the management of critical national infrastructure. This was not a minor back-office system. It was intended to form part of the operational brain of the railway network: the system through which train movements, control-centre operations, incident response, and national rail management were meant to be modernised. Instead, the public is now looking at years of delay, escalating cost, board-level concern, independent review, and a major write-down. Indra is not a small experimental supplier learning on the job. It describes itself as a global company operating across defence, aerospace, advanced digital technologies, mobility, and transport. That makes the reported failure more serious, not less. The question is no longer simply whether the software works. The question is who signed off the original assumptions, who monitored delivery, who escalated the risk, who protected the taxpayer, and why the warning signs were allowed to accumulate for so long. Ireland cannot keep treating large public technology failures as unavoidable weather events. They are not. They are governance failures. When a state transport technology project can burn through this level of public money without delivering the operational capability promised, the public deserves more than careful language, delayed reviews, and familiar reassurances. It deserves a full account: names, dates, decisions, milestones, warnings, consequences, and an explanation of why this was allowed to happen.

@KyoshiTakeshiro @SynthdataCo Thanks for the thoughtful read, Kyoshi. Probabilistic forecasting at the scale Synth is building requires a deep bench of competing models, which is exactly what the CrunchDAO network produces. Watch this space.

Compete here: hub.crunchdao.com/competitions/b…

Obesity ML Challenge Part 3 is launching today. A leading biomedical research center @Schmidt_Center is working on one of the harder open problems in obesity research. Instead of keeping it inside a closed in-house team, they opened it to a decentralized network of independent researchers all competing on the same question. The question: which gene pairs, out of 4,474,413 candidates, most strongly drive thermogenesis in fat cells, a leading path to treating obesity. The strongest submissions from Crunchers worldwide get tested directly in the lab.

🎉 The next phase of the @Schmidt_Center Obesity ML Competition is LIVE! hub.crunchdao.com/competitions/b… This challenge is open to researchers from all backgrounds — no machine learning expertise required.

@0xSammy @KhalaResearch Solid landscape. Good company to be in on that list!

I just published my 177th weekly newsletter on crypto AI & robotics This week provides a broader landscape analysis Which protocols are missing? Drop their core thesis in the comments below👇 Free to subscribe, link is in my bio; follow @KhalaResearch for more in depth reporting

We are welcoming @Graham_dePenros to Crunch as Strategic Advisor, leading institutional growth. Graham brings over three decades of experience operating where AI and emerging technology meet the real world. His career spans national defense and cybersecurity, healthcare and medtech, global enterprise logistics, and capital markets. He advises Tier 1 market intelligence research firms on emerging technology, and his research on cognitive security, AI governance, and data integrity is read closely by the agencies, firms, and capital allocators that put AI into environments where stakes are high. His work at Crunch is to bring non-consensus intelligence to institutional environments at global scale. Give a shoutout to Graham!

Louvre Palace. June 2-3. @herelle_jean is bringing the mining pool thesis to Proof of Talk. The same playbook that turned Marathon, Riot and CleanSpark into multi-billion dollar businesses, now applied to Bittensor. If you're in Paris, this is the talk.

@0xSammy Appreciate it, Sammy. Three rounds in on Structural Breaks with ADIA Lab and the questions only get harder, the signal only sharper.

@crunchDAO repeat institutional engagement on hard problems - fair play lads

Most models assume tomorrow behaves like yesterday. The moment that assumption breaks, every forecast built on it does too. Structural breaks sit inside billion-dollar automated systems, from market regimes in quant finance to climate sensors and healthcare diagnostics. Predicting them is what makes high-stakes automation robust. ADIA Lab is back with more work for the Crunchers, once again choosing an open and decentralized ML network, over a closed lab or Big Tech vendor. Not Palantir, not Quantum Black not McKinsey: Crunch. Let’s keep the great work Crunchers. Compete on CrunchDAO: hub.crunchdao.com/competitions/s…

@jumperz On the hardest problems, a global network of researchers competing head-to-head beats closed teams. We simply built the rails for institutions to access it and ADIA's renewal proves this model works.

ADIA could’ve gone to any big tech AI company instead they used a decentralized ML platform where people from all over the world compete to solve the problem wild tbh

$TAO live on @solana. Our crunchers have been mining @opentensor subnets for months, earning alpha tokens daily. With the bridge live, flipping those rewards just got frictionless. Two ecosystems we've been building on. LFG 🔥

The scientists behind our biomedical ML challenges sat down for a panel at the Broad Institute yesterday. @herelle_jean was up there with them. What it actually takes to build competitions like our IBD, cancer, and obesity challenges. What 1,000+ Crunchers from 62 countries iterating for months on these challenges actually produce. From the Eric and Wendy Schmidt Center's annual Biomedical Science and AI Symposium, here's the full video from the panel.

we never had such an opportunity to cure that many diseases. We need to throw more brains, more neural nets, and more compute into the biomedical space guys! 2 years since the Broad Institute has been using Crunch’s decentralized modeling network to improve supercharge their ML discovery and thousands of ML models later, Crunch feels like a core part of the lab. proud of the Crunchers. Big things ahead.

@Graham_dePenros @herelle_jean Thank you for the incredible writeup, Graham. Two days at the Eric and Wendy Schmidt Center at Broad Institute showed what becomes possible when decentralized collective intelligence sits at the same table as the world's best researchers. Full video coming soon.

Broad Institute and Crunch Lab Inc. (CrunchDAO) Panel Discussion on Machine Learning Challenges Day 2 has been fantastic and is wrapping up with Crunch CEO @herelle_jean and a stellar panel representing @crunchDAO Crunches results after detailed reports today and yesterday at the Eric and Wendy Schmidt Center at the Broad Institute of MIT and Harvard. Shootout to Andrew Zapo @zapoandrew and Gregoire Colcombet @GColcombet! Today's Crunchers showcased how they developed machine learning models with reports today's from multiple Crunches including: • Biological data in the age of AI - Michael Bronstein, University of Oxford; Aithyra Institute • Interpretable Representation Learning for Wound Healing Dynamics - Bianca Dumitrascu, Columbia University • Generative Design of Cell Type-Specific RNA Splicing Elements for Programmable Gene Regulation - Maile Hirschmann (Jim), Fei Chen Lab, Harvard University/Broad Institute • AI-Enabled Phenotypic Profiling of Parasite Infection Using Optical Pooled Screening - Eden Yifrach, Lourido Lab, Whitehead Institute • LabOS: The AI-XR Co-Scientist That Sees and Works With Humans - Mengdi Wang, Princeton University • Algorithms for Biomolecular Structure Determination at the Proteome Scale - Ellen Zhong, Princeton University • AI-driven design of anticipatory vaccines - Navami Jain, Debora Marks, Harvard Medical School • Three impactful problems in drug discovery that might be solvable with machine learning (and what we are doing about them at Relation) - Lindsay Edwards, Relation Therapeutics • Uncovering Dark Metabolites through Neural Structural Elucidation with Tandem Mass Spectrometry - Runzhong Wang, Coley Group, MIT • Decoupling and Dimensionality: Two Frameworks for Interpretable Multi-Modal Representation Learning - Viktoria Schuster and Sana Tonekaboni, Uhler Lab, MIT • ProtGPT3: an Open-source family of Promptable and Aligned Protein Language Models - Michele Garibbo, Centre for Genomic Regulation • Panel discussion on machine learning challenges

The Eric and Wendy Schmidt Center at the Broad Institute is running its annual Biomedical Science and AI Symposium today and tomorrow in Cambridge. Two days of researchers from MIT, Harvard, and beyond on the intersection of ML and biology. Free livestream, open to all. Tomorrow at 4 PM EDT, our founder Jean Herelle (@herelle_jean) joins a panel on machine learning challenges. Tune in: broad.io/watch-ewsc-sym…