stephen tse

@harmonyprotocol founder, compiler phd, google+apple map

[stephen tse @stse]

https://t.co/zpEMFOBJS6

[Harmony 💙 @harmonyprotocol]

🏀 https://t.co/XdBeQyMr40

Dunk like it's the finals.

[stephen tse @stse]

"a 90,500-square-foot, 16-floor building on the corner of Sixth and Market for only $11 million.

two floors from the Spaceship: fourth-floor robotics wing, headed by Xenia and Vitaly Bulatov

Carbon Origins, which develops VR remote tools for robots, and Robonomics, whose humanoid is housed here - the “Cyberpunk Lab.”

10 minutes away from Union Square, The Bulatovs keep a $50,000 Unitree Robotics Humanoid G1 and a $75,000 Boston Dynamics Spot dog

https://t.co/ArQmSHmMjb

[Vitaly Bulatov @vitl2907]

Spot and I visited @frontiertower - a vertical village for frontier tech in SF - a perfect location for our Cyberpunk Lab! There we plan to collaborate with researchers, artists and developers on giving agency to #HumanoidRobots.

Building a strong community, partnering with top talent on AI model testing, exploring the human <> robot interaction through art and science — are the activities that will help us create a secure, fair and thriving economy of the 21st century. One where robots are a major part of economic activity and wealth creation. More details to come soon, stay tuned!

PS Spot had a fun time playing with @VitalikButerin 🐱🐶🤖
Would you have a robodog as a pet?

[stephen tse @stse]

https://t.co/poSX1QSIzL

https://t.co/M2MVSh2XZp

[Elon Musk @elonmusk]

I have come to the perhaps obvious conclusion that accelerating GDP growth is essential.

@DOGE has and will do great work to postpone the day of bankruptcy of America, but the profligacy of government means that only radical improvements in productivity can save our country.

[stephen tse @stse]

"The chart shows a slowdown in total factor productivity (TFP) growth after 1971, when the U.S. ended the Bretton Woods system, decoupling the dollar from gold. This shift to fiat currency, combined with 1970s oil crises and stagflation, likely reduced business investment, impacting TFP.

A temporary IT-driven boost occurred in the 1990s, but growth slowed again. Some argue measurement issues, like undervaluing digital services, or sectoral shifts to services exaggerate the slowdown. Others suggest declining efficiency, not innovation, is key.

[Marc Andreessen 🇺🇸 @pmarca]

"What happened in 1971?" update. HT @MargRev

[stephen tse @stse]

“Sacktor and Fenton’s star protein PKMζ gained widespread attention, and labs around the world found that blocking it could erase various types of memories, including those related to fear and taste. PKMζ seemed like a sweeping explanation for how memories form and are maintained at the molecular level.

the interaction between KIBRA and PKMζ is crucial not only for forming memories, but also for keeping them intact over time.

[Albert Einstein @AlbertEinstein]

Individually, two particular proteins in the brain don’t last a lifetime — but by binding to each other, they help ensure your memories might. https://t.co/uiRtGV6pH5

[stephen tse @stse]

“Then there’s Apple’s push into robotics, which will include a tabletop machine with a robotic arm (building on the company’s still-pending plans for a smart home hub).

That would feature an AI assistant with its own personality.

[Evan @StockMKTNewz]

Apple $AAPL is planning a big 2027 with foldable all-screen iPhones, smart glasses, camera equipped wearables, table-top robots and more - Bloomberg’s Mark Gurman

[stephen tse @stse]

"A cornerstone of human dexterity lies in the thumb, the distinctive mobility of which allows it to pair with any other finger.

This enables fine prehension, precise handling of tools, and the rotational manipulation of objects held within the hand. Nevertheless, actions without perception are as meaningless as a skeleton without muscles.

The evolutionary emergence of soft tissues, including muscles, tendons, and highly sensitive skin, was equally fundamental to human dexterity.

https://t.co/DW2MeuU4c6

[Silicon Valley Story @theSVstory]

In good hands: A case for improving robotic dexterity - Science - https://t.co/n1yWA2Crqj #robots #Robotics

[stephen tse @stse]

“And we couldn’t just teach Vulcan with computer simulations, but trained its AI on physical data that incorporates touch and force feedback. It tackled thousands of real-world examples, from picking up socks to moving fragile electronics.

Vulcan even learns from its own failures, figuring out how different objects behave when touched and steadily building up an understanding of the physical world, just like kids do.

[Werner Vogels @Werner]

Meet Vulcan: Amazon's first robot with a sense of touch. https://t.co/roGUfJbuW6 #AI #robotics

[stephen tse @stse]

"Consensus relies on a 2-of-3 MPC threshold signature scheme (TSS), requiring a majority of Guardians to sign any critical protocol event. During distributed key generation, each Guardian receives an independent share of the private key, which remains encrypted at rest using KMS services and is only combined at runtime for signing in the secure enclaves. These enclaves keep key material isolated from the broader system, preserving confidentiality even if one Guardian’s environment is compromised.

https://t.co/MMHzsSAgCj

https://t.co/dsLQXOkDf1

https://t.co/9YFQC7cRuX

[Picasso @0xPicasso]

My complete modular HyperEVM farming blueprint:
Been deep in hEVM for months & got DMs on what my strategy was. So here’s my full framework. No alpha gatekeeping. Just what’s working for me.

I personally mix all these approaches depending on market conditions.
I've rated each strategy on an "offensive scale"
(0% = super safe, 100% = full degen mode).

I painted this graph so it helps you build your own strategy. I share mine, but use it flexibly to design what fits you best.

💧 Strategy A: The Safety Net (20% Offensive)
This is my go-to when I want steady yields with minimal risk. I split it like this:

> I deposit 20% of my stack on @sentimentxyz from which ½ goes in a leveraged position. The other ½, I borrow wHYPE against wstHYPE that I've swapped on @ValantisLabs.

> Then with the borrowed wHYPE, I split it in 2 and lend on Hyperlend and @hypurrfi for more HYPE again.

>Finally, I take that fresh HYPE and deposit it on @ValantisLabs to complete the loop. Or you can also deposit some in the @harmonixfi testnet HYPE vault, which is currently capped to only 30 HYPE.

Why this works: Low risk of liquidation while still earning solid yields. Perfect entry point if you're looking to earn some yield on your HYPE or just want to sleep well at night.

💧 Strategy B: The Balanced Approach (N/A% Offensive)

This is where I put 80% of my capital - balancing decent yields with manageable risk:

> A little over ½ goes on @felixprotocol stable farm, which I mint feUSD with a health factor > 2.5. Always monitor the redemption rate too. (learned the hard way.)
Since they've launched the vanilla market, I'm now depositing some HYPE for USDe as well.

> The other portion went on @hyperlendx from which I borrowed $HYPE that I deposited on @hyperbeat. Really bullish on this one, the recent surge in TVL is crazy, and imo the team is solid, love the way they manage their business. Recntly increased my allocation by depositing in the USDT0 vault.

What I'm seeing: This strategy is to me the base that allows me to leverage HYPE all over the whole ecosystem, as you can see in the graphic.

💧 Strategy C: The Degen Plays (75% Offensive)
NGL, this is the fun part. I only use about 15% of my capital here. Here you use your stables and go long on other assets (or just more HYPE)

> So, I bought some HYPE with the borrowed stables, sent very few on @HanaNetwork, bought some keys on @fanappX (a project that I like and find underrated). Team is building a better friendtech version, and community is big time sleeping on it. Early users are being rewarded. Cant wait for the trading feature to be deployed.

> Never been a solid trencher, I still sent some HYPE on @LiquidLaunchHL. As a big liquidswap bull, I bought a nice bag of $LIQD the best Hyperliquid beta imo.

> Since @okto_web3 integrated the Hypezone, I opened a new wallet onto which i sent some $HYPE and explore the dapp. Gotta admit, the wallet is easy to use and brought me to discover @LiquinaHL, a project i wasn't really familiar with.

>Made some good money on @harborfinance_ as well, that allowed me to pay back many of my loans.

>For the NFT degens, you can also buy them on @DripTrade or even @LootLiquid once they launch.

Real talk: This strategy is really for the ones who like to gamble. It has been the source of my biggest wins lately. You could also send some stables on @okto_web3 or @HyperliquidX and open some trades too.

💧 Strategy D: The Stablecoin Maximizer (5% Offensive)
When I want to play it safer but still earn solid yields:

>With the remaining stables, I provided a lot in LPs during the DEX farming season and lately to the USDE/feUSD on @CurveFinance. Also deposited a good amount in the USDT0 vault on @hyperbeat.

>Sent also a bunch on HLP and @harmonixfi on which you can benefit from 15% apy on your USDC. But overall, can't wait to deposit some stables whenever @liminalmoney sends out more ref codes.

>Been testing @hyperdrivedefi on testnet, waiting for mainnet to boost my HLP allocation - to consistently get 10-30% on stables on autopilot.

Bonus move: I'm also farming @unitxyz by bridging some BTC and ETH on hypercore. Lately been enjoying depositing some uETH on @KeikoFinance to maximize my farming options with ETH on hyperEVM.

What's your HYPE farming strategy looking like?
Drop it below and let's compare. Always looking for new alpha. My dms are always open.

NFA. DYOR and adapt to your own risk tolerance. Any strategy can go to 0 if shit hits the fan.

[stephen tse @stse]

"RUKA features 15 degrees of freedom across its five fingers, driven by 11 Dynamixel servo motors housed in the forearm to save space and manage heat.

The finger joints are coupled—one motor flexes multiple joints via tendons made from high-tensile braided fishing line running through PTFE tubing. Passive extension is achieved using embedded springs, further simplifying the actuation.

https://t.co/DC5fJgB7TZ

[Irmak Guzey @irmakkguzey]

Despite great advances in learning dexterity, hardware remains a major bottleneck. Most dexterous hands are either bulky, weak or expensive.
I’m thrilled to present the RUKA Hand — a powerful, accessible research tool for dexterous manipulation that overcomes these limitations!

[stephen tse @stse]

"The using declaration ensures that readerResource[Symbol.dispose]() is called when the code block exits, and remembering to call releaseLock is no longer needed because the using declaration handles it.

Integration of [Symbol.dispose] and [Symbol.asyncDispose] in web APIs like streams may happen in the future, so developers do not have to write the manual wrapper object.

[JSWORLD Conference @JSworldconf]

“JavaScript’s New Superpower: Explicit Resource Management” Introduces a new JavaScript feature that simplifies and automates the cleanup of resources like file handles, network connections, and stream readers.

👉 https://t.co/WxLT5V30Nn

[stephen tse @stse]

"The V2 iteration specifically features upgraded arms and hands.

Each arm offers 6 DoF, while the three-fingered end-effectors provide 7 DoF, featuring force control and optional touch sensing for adaptive grasping.

[Humanoids daily @humanoidsdaily]

For more details: 👇📗👀
https://t.co/8RSP68tjl9

[stephen tse @stse]

“took a summer job writing a COBOL compiler”.. his early programming experiences spanned IBM mainframes with PL/1 and Fortran, PDP-8 assembly and CDC 6400 code

“the first 90% is easy to figure out, and the last 10% is just super hard”: the challenge with features like generics and lambdas was determining the best implementation approach

“The Java storage management has been more efficient than the malloc, than the C storage management for really long, but now it’s just stunning.” Garbage collection pauses that once took “10 or 20 seconds” can now be reduced to milliseconds with careful tuning

[Andrés-Leonardo Martínez-Ortiz, PhD @davilagrau]

Java at 30: The Genius Behind the Code That Changed Tech
#java #people
https://t.co/FmW8PwPeap

[stephen tse @stse]

"Dexterity: The Final Frontier in Humanoid Robotics, Tackled by an Expert Gathering of Minds

Rich Walker, Director, Shadow Robot Company; Rob Knight, Founder & CEO, The Robot Studio; Jay Li, Co-Founder and CEO, Proception; Stefan Weirich, Co-Founder and CEO, Mimic Robotics; Aadeel Akhtar, Founder and CEO, Psyonic; Francesca Cini, Team Leader Robotics, Prensilia; Harpal Mandaher, CEO, Sarcomere Dynamics.

[Tom Anderson, Principal Analyst & Co-founder, STIQ.]

[clem 🤗 @ClementDelangue]

How would a 🤗 look like? From https://t.co/jzGDnLk7ek

[stephen tse @stse]

“I am reminded of a time, three decades ago, when I emigrated to America. As a _, I was drawn to the exhilarating and innocent optimism of Silicon Valley, to collaborate with people driven to create amazing products that elevate humanity.”

He then co-founded io last year with Apple alumni Scott Cannon, Evans Hankey and Tang Tan. Hankey was Ive’s successor at Apple and remained at the company until 2023, while Tan led iPhone and Apple Watch product design until 2024. Cannon worked at Apple before co-creating the once-popular email app Mailbox, which was acquired by Dropbox Inc.

https://t.co/scoAK6Z06W

[stephen tse @stse]

"each of the fingers has a flexible reflective membrane beneath its 7-mm to 8-mm skin. Beneath that is an array of LED lights and phototdiodes. Each of the lights is cycled on and off for less than a millisecond, and the photodiodes record the reflection of the light off the inner membrane.

“When a finger touches anything, it bends the ‘skin’ and changes how the light bounces around,” which then tells the digit how tightly it’s gripping the object, and exactly what part of the finger’s surface is touching the object

[ASME.org @ASMEdotorg]

This groundbreaking robotic hand can manipulate objects without relying on vision, thanks to advanced touch sensing with motor-learning algorithms: https://t.co/ftuGs9CR3D

[stephen tse @stse]

"traversing the reasoning manifold..

.. curate reasoning traces off that RL model to fine-tune a completely different second model

Deepseek was less about RL and more about generating a lot of really high quality reasoning traces

[Mehmet Burak Sayıcı @BurakSayici]

Deep Learning is Applied Topology:
Everything lives on a manifold https://t.co/QH3taGYWFJ

[stephen tse @stse]

"the original PINN paper has a whopping 14,000 citations, making it the most cited numerical methods paper of the 21st century.. becoming the second most-cited numerical methods paper of all time

My lab uses PDEs to model the behavior of plasmas, such as inside fusion reactors and in the interstellar medium of outer space.

Paper after paper haddemonstrated the ability to solve PDEs faster—often orders of magnitude faster—than standard numerical methods. The examples that excited my advisor and me the most were PDEs from fluid mechanics, such as the Navier-Stokes equations.

[Nick McGreivy @NMcGreivy]

In a guest post for Understanding AI (@binarybits), I write about how I got fooled by AI-for-science hype, and what it taught me.

I argue that AI is unlikely to revolutionize science, and much more likely to be a normal tool of incremental, uneven scientific progress.

[stephen tse @stse]

Compiling OCaml to the TI-84+ CE Calculator

"replace all local variables with explicit reads and writes into a global stack. That way, we can scan the global stack to find all of our live objects when it comes time to GC.

https://t.co/diLeoz3div

[Nathan Farlow @0x1337cafe]

Published two posts about some art which I've made. This first piece is titled "When you read this I'll be gone". It incorporates time locked crypto! https://t.co/VJYObncn5f

[stephen tse @stse]

"their browser will spawn dozens of pop-up pages that bounce around the screen, all of them featuring information on how to defect from North Korea.

Then loud music plays—a rickroll, “The Star-Spangled Banner”—before the computer starts downloading random files and emits an ear-splitting beep.

[Evan Kirstel #B2B #TechFluencer @EvanKirstel]

Your Favorite New Coworker Is an AI-Enhanced Operative From North Korea

For years, North Korea has been secretly placing young IT workers inside Western companies. With AI, their schemes are now more devious—and effective—than ever.

https://t.co/GFfrAKkTjP #RSAC