𝑻𝒉𝒐𝒎𝒂𝒔 🧢

Quant trading is solved. Introducing: Minara Strategy Studio. Design strategies, backtest, compare PnL, avoid brutal drawdowns, optimize parameters, go live, all in plain words. Join the closed beta: strategy.minara.ai 🧵

The moment AI trading truly matures might look like the invention of the stock exchange: most people didn't understand what it was, then gradually, they trusted it with their money.

Can a trader who only knows EMA build a Sharpe >1 strategy in a few minutes? The answer is yes and this article showed every iteration.

: : We're excited to announce that @FourPillarsFP has raised Series A Funding from @PanteraCapital and @further. This round marks our next chapter: scaling from a research firm into an institutional-focused blockchain research and infrastructure provider.

what if the answer to everything was all of us hitting one at the same time? we wouldn’t really know for sure until we tried it just asking questions

The market clearly would rather be hard.

NEW GPT, CHINA STEALS MYTHOS?, MARK HALPERIN x.com/i/broadcasts/1…

Minara AI Strategies are now live to deploy. 📈 And here comes the Minara Strategy Studio Content Campaign. 🔮 If you’re having fun with Strategy Studio strategy.minara.ai , please share what you’re building with: - Fantastic strategies - Solid backtests - Historical performance or, any cool stuff you built with @minara products 🪄 Rewards: - 100 sparks for genuine posts 🔥 - 500 sparks for articles or threads 🧵

A startup called Kos.AI is targeting a hidden bottleneck in AI infrastructure: CFOs struggling to manage massive payments and paperwork. Full story in today's AI Infrastructure newsletter: thein.fo/4cGTloZ

To accelerate physical AI development, we need “the ability to orchestrate perception, reasoning, and action into sustained, autonomous, closed-loop systems that operate over long time horizons without human intervention.” Antioch is building the architecture for this vision — an agentic dev environment for physical systems, with simulation as the runtime. @oyhsu laid out the emerging physical AI paradigm in a piece for a16z: a16z.news/p/frontier-sys…

⚛️ EquiformerV3 is here ! State-of-the-art on Matbench Discovery and OC20🏆 Better scaling, improved sample efficiency, stronger generalization. The Equiformer line keeps getting better! Huge congratulations to Yi-Lun Liao and Tess Schmidt for driving this vision so brilliantly ! It's been inspiring to watch it come to life at this level. Having witnessed & used previous Equiformer versions, I am grateful to have contributed to this one, even if it was a small piece of a much larger puzzle. Go check it out 🤗

Full self-driving mode is finally here 🏎 🦅 Introducing Falconer Update: automatic updates for your docs. Toggle it on to auto-update. When a PR merges, Falconer reads the diff, keeps the docs in sync, and pings you in Slack. Just keep shipping. The docs will keep up.

We’re building a community of the greatest minds in RL research. This week, we kicked off our first private dinner in SF for handpicked researchers working on autonomously scaling RL environments. Closed-door technical conversations only, no VCs. All people who are betting their lives that: 1) The data problem is the biggest bottleneck in model advancement today 2) We can collectively solve the problem by autoscaling this data instead of relying so much on manual human input We’ll be hosting these hyper-specific research-focused events at great restaurants on a regular basis. If you’re working on these problems: - Autonomous scaling of environments for RL post-training - Unsupervised environment design in these domains: - Computer/browser-use agents - Long-horizon tool-use enterprise agentic workflows - Autonomous AI research we'd love to have you.

A first look at what we’re building at Sudo AI. Introducing #sudo R1. We train robots entirely in simulation, then deploy them directly to hardware - no data collection, per-object tuning, or calibration. In this first release shared in our blog, #sudo R1 runs 60 minutes uncut and reliably picks objects it has never seen before: glass, fabric, metal, irregular shapes - under changing lighting, clutter, and physical interference. Simulation-trained systems rarely hold up in practice. We think that’s starting to change. If you’re building in robotics or thinking about where this is going, follow along at sudo.ai

Today we're pleased to share that Sygaldry has raised $139M to build quantum-accelerated AI servers for AI data centers, delivering more compute per watt. We're hiring brilliant, curious, kind, and collaborative scientists, engineers, and more. bit.ly/48KBufk

We're proud to have collaborated with @nvidia and @QCHardware on the release of NVIDIA Ising Calibration, the first open Vision Language Model for quantum calibration. Together with EeroQ, we used it to build a working proof of concept for an autonomous quantum computing lab. 🧵

Welcome to MTS | April 20 x.com/i/broadcasts/1…

A suite of cards, a spread and a decision

we're hosting a creator hack on Friday in SF. 10-15 creatives will have 12 hours to make a film. on Saturday, they will present their work in front of a panel - announcing our first judge today: > @adilmania from @siliconmania <