Articles by Martin
-
Mid-Senior UI UX in Sacramento
Mid-Senior UI UX in Sacramento
A recruter friend of mine is looking UI UX person in Sacramento 50-60 an hour, full time through the end of the year…
Activity
663 followers
-
Martin Stack reposted thisMartin Stack reposted thisFor my career, I studied people, environments, and decision-making. Somewhere along the way I apparently made the completely rational decision to write doctrine. I did not have that in the 2026 bingo card... Today I'm publishing the first public white paper introducing Residential Intelligence. This isn't a marketing piece. It isn't a sales brochure. It's my attempt to put structure and language around something I've believed for a long time: We understand homes. We understand security. We understand crime. But we don't have a discipline dedicated to understanding the residential environment itself. Residential Intelligence starts with a different question: Do we truly understand the environment in which we live? This paper is the beginning of that conversation. It lays out the foundational thinking behind the discipline and the philosophy that drives the work we're building at OutThink the Threat. https://lnkd.in/gxNZ-siQ If you're in intelligence, law enforcement, executive protection, community management, urban planning, CPTED, emergency management, or simply enjoy challenging conventional thinking, I'd genuinely appreciate your feedback. Agree with it. Disagree with it. Tell me where the reasoning holds and where it doesn't. Good ideas survive scrutiny. The white paper is available now as a free download on my website. No, it’s not gated. No, I don’t want your email. Sometimes the best way to outthink a threat isn't buying another camera. Sometimes it's learning to understand the environment you've already been living in. #ResidentialIntelligence #HomeVulnerabilityAssessment #Intelligence #RiskAnalysis #CommunitySafety #ExecutiveProtection #ProtectiveIntelligence #CriticalThinking #UrbanPlanning #HOALeadership #Security #OutThinkTheThreat #DeusXDefenseOutthink The Threat| Intelligence-Driven Security AssessmentsOutthink The Threat| Intelligence-Driven Security Assessments
-
Martin Stack shared thisOutstanding! The present is AI on Silicon, the future is AI on Optical. This is wonderful news.Martin Stack shared thisTogether with Mr. Young Sohn (Co-Founder of Walden Catalyst Ventures), SK Group, Chunghwa Telecom Co, Ltd., and Development Bank of Japan Inc., we are establishing the IOWN AI Fund, expected to launch this month with a target size of $500 million. The fund aims at building the IOWN ecosystem and creating new business opportunities through investments in cutting-edge technologies for the AI era. #NTT #AI #semiconductors #photonics #SiliconValley
-
Martin Stack shared this🤩GitHub - backnotprop/plannotator: Annotate and review coding agent plans and code diffs visually, share with your team, send feedback to agents with one click.GitHub - backnotprop/plannotator: Annotate and review coding agent plans and code diffs visually, share with your team, send feedback to agents with one click.
-
Martin Stack reposted thisMartin Stack reposted thisNVIDIA GPUs with GDDR6 memory can be used to take full control of a system, including a root shell, bypassing hardware defenses that were supposed to stop exactly this. Three independent research teams published GPU attack research at the same time. One of them goes further than anything before it. There is currently no fix for consumer GPUs. 😏 The attack is called GPUBreach, and it comes from researchers at the University of Toronto. To understand why it matters, you need to know what Rowhammer is. DRAM memory is built from rows of tiny cells packed extremely close together. When you hammer the same row with repeated memory accesses, the heat and electrical noise leaks into the rows next door and flips bits that nobody was supposed to touch. A 1 becomes a 0. A 0 becomes a 1. In memory that does not belong to you. GPUBreach takes this further than anyone has before. Where earlier research caused data corruption, GPUBreach causes full privilege escalation. The attack corrupts GPU page tables, basically the address book of the GPU that keeps one program from reading another program's data. Once those are compromised, an unprivileged CUDA kernel has arbitrary read and write access across the entire GPU memory space. But GPUBreach does not stop at the GPU. The standard defense here is IOMMU, which restricts what memory a GPU can access. NVIDIA, AMD, and Microsoft all recommend keeping it enabled. The two concurrent research papers, GDDRHammer and GeForge, both require IOMMU to be disabled to reach the CPU. GPUBreach does not try to go around IOMMU. It goes through the NVIDIA driver instead. → Secret post-quantum cryptographic keys can be stolen from NVIDIA's cuPQC library during active key exchanges. → LLM weights can be leaked from GPU memory — billions in training costs, gone. → AI model accuracy can be silently destroyed from 80% to zero with no error messages and no crash. → Full root shell on the host, with IOMMU enabled. The University of Toronto researchers reported this to NVIDIA, Google, AWS, and Microsoft on November 11, 2025. That is almost five months ago. Google paid a bug bounty of $600. For context, Google pays up to $250,000 for a critical Chrome sandbox escape. NVIDIA is worth close to a trillion dollars. Google is worth over two trillion. And the researchers who spent months finding this walk away with $600. It is not a reward. It is an insult. Understanding how attackers escalate from a foothold to full system control is exactly what I cover in my ethical hacking course. → https://lnkd.in/e6KECVJc Research & writing: Jolanda de Koff Sharing is fine. Copying without credit is not. Unfortunately, the full story doesn't fit in a LinkedIn post. → https://lnkd.in/eg89kSMU #EthicalHacking #GPUBreach #NVIDIA #Rowhammer #CyberSecurity #InfoSec #PrivilegeEscalation #GPU #GDDR6
-
Martin Stack reposted thisMartin Stack reposted this💣 Two almost simultaneous relevant papers on #quantum #cryptoanalysis. 👉 "Shor’s algorithm is possible with as few as 10,000 reconfigurable atomic qubits" (https://lnkd.in/eyGiqXQt): This document, supported by trusted names like John Preskill, discusses advances in error-correcting codes and other efficiencies that could be leveraged in neutral atoms quantum computers. They discuss attacks on RSA using as few as 10,000 atomic qubits, although at a great cost in time. Their most time-efficient architectures can enable run times of 10 days for ECC–256 with ≈26,000 qubits, and 97 days for RSA–2048 with ≈102,000 qubits. See the graph below. 👉 "Securing Elliptic Curve Cryptocurrencies against Quantum Vulnerabilities: Resource Estimates and Mitigations" (https://lnkd.in/e_HsxUcx, https://lnkd.in/eakjd4HU): This paper has been published by Google Research and counts also with trusted authors from Google, Ethereum Foundation, University of California, Berkeley and Stanford University, like Craig Gidney, Justin Drake, or Dan Boneh. The paper is a comprehensive review of #quantum #security in #blockchain that deserves a careful reading. They demonstrate that Shor’s algorithm for breaking 256-bit ECC can execute with either ≤ 1200 logical qubits and ≤ 90M Toffoli gates or ≤ 1450 logical qubits and ≤ 70M Toffoli gates. On superconducting architectures with 10^−3 physical error rates, it could be executed in minutes using <0.5M physical qubits. They analyze how this can enable different attack scenarios to cryptocurrencies. 👉 This not a sudden breakthrough, but steady, credible progress in quantum cryptoanalysis. 💡What stands out is not just feasibility, but implications. 🚩 Although substantial expertise, experimental development effort, and architectural design are required, quantum systems capable of breaking today’s cryptography are not speculative. This underscores the importance of ongoing efforts to transition widely-deployed cryptographic systems toward post-quantum standards. 🚩 The emergence of CRQCs represents a serious threat to cryptocurrencies. ✏️ The Bitcoin community needs to face urgent and difficult decisions regarding legacy assets, such as the 1.7 million bitcoin locked in P2PK scripts and an even greater amount of assets vulnerable due to address reuse. ✏️ Ethereum is more exposed than Bitcoin due to the prevalence of at-rest vulnerabilities, but its recent active steps towards PQC migration promise a more expedient transition to quantum-safe protocols. This is critical since the tokenization of real-world assets is expected to open up markets projected to exceed 16 trillion USD by 2030, breaking the “too-big-to-fail” economic stability thresholds. ✏️ There is time to migrate public blockchains to PQC, though the margin for error is increasingly narrow.
-
Martin Stack reposted thisGoogle just cut the quantum threat to Bitcoin by 20x. New Google Quantum AI research: breaking the cryptography securing Bitcoin and Ethereum may need fewer than 500,000 qubits down from millions. Google's best chip today has 105. The gap is closing faster than anyone projected. But this was never just a crypto problem. The Elliptic Curve Discrete Logarithm Problem secures TLS, digital signatures, and authentication systems across every sector. Blockchains are just the most exposed: smaller keys, public transactions, no recourse, millions of vulnerable addresses already on chain forever. So what's the fix? PQC (Post Quantum Cryptography) and Quantum Cryptography are not the same thing. QKD uses quantum mechanics itself to secure communication. Theoretically unbreakable. Practically unusable at scale, it needs dedicated hardware and fiber, and tops out at point-to-point. PQC is deployable today. But it's still just math on classical computers. "Hard for quantum computers" is an assumption, not a proof. One NIST finalist was broken in an hour on a laptop. We're not solving the problem. We're buying time with a different bet. Now layer in agentic AI and the stakes multiply. These systems don't just answer questions. They authenticate, sign requests, hold keys, and execute decisions autonomously across your entire stack. No human in the loop. No second look. Every cryptographic weakness in your infrastructure is inherited by every agent running on top of it at machine speed, at scale. A quantum capable adversary doesn't need to breach your perimeter. They compromise the identity layer your agents trust. Silently. Completely. That's the feature of agentic AI. It's also the attack surface. What this means right now: → Crypto agility isn't a roadmap item, it's a survival requirement → "We deployed CRYSTALS Kyber" is not a finished sentence → Your AI agent trust model belongs in your post-quantum threat assessment → The window to migrate deliberately is open. It won't stay open. Have you mapped your agentic systems to your cryptographic risk surface? #Cybersecurity #QuantumComputing #PostQuantum #Cryptography #CISO #CryptoAgility #AgenticAI #QSVA
-
Martin Stack shared thisExcellent !😎Martin Stack shared thisFrom collaborative 3D design to real-time assistance on the manufacturing floor — the opportunities for XR are endless. 🥽💼 goo.gle/androidxr We’re excited to launch Android Enterprise management for Android XR to help businesses deploy XR experiences at scale! You can now deploy and secure Samsung Mobile Galaxy XR powered by Snapdragon using the EMMs you already trust, like: ArborXR, ManageXR, Microsoft Intune, Omnissa, Workspace ONE, Samsung Knox Manage, and SOTI MobiControl. #AndroidXR #AndroidEnterprise
-
Martin Stack reposted thisMartin Stack reposted thisWe invite our community to join us in celebrating the life and legacy of Tom Matano. On Thursday, April 2, President Stephens will host a Celebration of Life at the Academy of Art University Automobile Museum (1849 Washington St.) from 5:30–8:00 p.m. Tom’s impact on the Academy’s School of Industrial Design, and on all who had the privilege of learning from and working alongside him, was truly indelible. We hope you’ll join us as we honor his remarkable career and lasting contributions. 🤍
-
Martin Stack reposted thisMartin Stack reposted thisYes OpenClaw has security flaws, big ones. But if you think that's gonna slow this train down, you are gonna get run over. This rocket is only accelerating. I've used OpenClaw and had it running both Claude Code, Codex with multiple instances each working all night. No fancy GSD, Gas Town or other framework. Just got my project going with 1:1 Claude Code, then hooked up OpenClaw to Telegram and had it push these elite level coders all night. What we think of as software today is transitory. It's not a law of nature. Programmer wasn't a job a few short years ago, and it soon won't be again. Systems thinking and engineering are very much alive, but the lines between compiled code, interpreted code, and inference are going to blur and then vanish. The computer use model will continue to eat all other software, burying it down into the stack out of the reach of human hands and eyes. AI will be used to secure AI. There is 0% chance that humans will keep up in this race. We must, and we will move up the stack. There's a gigantic catch 22 for big companies right now. It's going to cause a seismic shift in and roll over in the fortune 500 and 5,000 or maybe even down into the broader business landscape. All of these companies have too much legal liability to fully embrace the AI wild west. They can't unleash Claude Code or OpenAI Codex into their ecosystem. The attack vector and risk is way too high. Prompt injection, misaligned models, and the like will make this a no-go for true adoption. Individuals and small teams will YOLO it on these tools. They have everything to gain and very little exposure to the risks. They can worry about the security later when they have something worth protecting. The security paradigm for AI will improve. The models will align, the layers of alignment and checks and balances will emerge. But by the time this is all figured out, these small teams are going to absolutely cook (as the kids say) the old guard. I'm guessing the physical world companies will quickly catch on to this game. They will outsource the risk to the startups. The SaaS-pocolypse will look like a picnic compared to the absolute implosion as the broad retail investment market wakes up to this. Startups have always won against big entrenched players because of these dynamics, but this iteration is going to be like nothing we've ever seen before.
-
Martin Stack liked thisAfter sometime working with AI assistants, we found how powerful they are for us (personal and professionally) and our close relatives. However, installing, operating, and securing it is not an easy task. That´s why we created HolaClaw, a native macOS application to install and operate OpenClaw. Configure your provider, a channel to talk with it, and your assistant is ready in 5 minutes. A writer assistant, study buddy, travel agency, or a elderly compain. You name it, shape it, and control it. Get it on https://holaclaw.ai 👋Martin Stack liked thisHolaClaw is the safe and easy way to run OpenClaw on your Mac. Just download, double-click, and a few minutes later you have your own, personal digital assistant up and running. Everything runs in a separate virtual machine, keeping your system secure. Give it a try today: https://holaclaw.ai Why did we build HolaClaw? At Endor we have been early adopters of AI and have built some cool open source tooling for agentic coding. OpenClaw is an extremely popular personal digital assistant that has been enthusiastically adopted by many developers and entrepreneurs (including us!). HolaClaw makes it accessible to a wider audience. We encapsulate the complexity so the end user does not deal with installing runtimes and dependencies from the command line, just a regular GUI Mac application that can be launched. Similarly, though OpenClaw has made great strides in the security arena since its launch, we go one step further, running each agent in its own fully isolated environment, and focusing on making the installation default-secure yet as easy as possible. HolaClaw works with your existing provider, whether it is Anthropic, OpenAI, Grok, Open Router and so on. If you have good hardware, HolaClaw’s ability to automatically install and run fully local models provides an interesting alternative for users that are extra-conscious about privacy and cost You can download and run it today by visiting https://holaclaw.ai Let us know your experience!
-
Martin Stack liked thisMartin Stack liked thisLast week, I wrapped up an incredible four years at Toast. It’s hard to summarize what these four years have meant to me. I had the chance to work with some of the smartest, kindest, and most passionate people I’ve ever met, not just within Design but across the company. Every day I got to learn from people who cared deeply about our craft, our customers, and each other. I also had the privilege of building for restaurants and small businesses. Spending time with customers, hearing their stories, and seeing the impact our work had on their businesses is something I’ll always carry with me. Many of those customers became friends along the way. I’m grateful to Aman, Steve, and many more at Toast for giving me opportunities that shaped both my career and how I lead. Thank you as well to the incredible teammates, partners, and leaders I had the chance to work alongside and occasionally have some of the best debates over dinner with. There are too many people to name, but you’ve each left your mark on me. More than anything, Toast reenforced to me the belief that the best products come from teams that care deeply about the people they’re building for. Leaving is bittersweet. I’ll miss the people, the customers, and the mission more than I can put into words. I will for sure continue to ask every restaurant I walk in about Toast and share back that feedback! I’m excited for what’s next, and I can’t wait to share more soon.
-
Martin Stack liked thisMartin Stack liked thisFall 2026 my gonks 😌 New story, the same epic neon drenched ass whooping, in a dystopian setting 🔥
-
Martin Stack liked thisMartin Stack liked thisWe’ve designed and built our first AI chip: Jalapeño. Designed from the ground up by OpenAI and brought to production with Broadcom, Jalapeño is purpose-built for the LLM workloads powering ChatGPT, Codex, the API, and future agentic products. Chips are foundational to the AI economy. Building our own expands our full-stack platform from products to models to infrastructure, and will help us scale intelligence, serve more people, and expand access to AI. https://lnkd.in/gg8UvTwE
Experience
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
-
Education
-
-
-
-
Other similar profiles
Explore more posts
-
Jesse Landry
Podonos just raised $2.4 million in pre-seed funding, and the signal is loud and clear: the voice AI industry finally has someone measuring the sound, not just making the noise. Based in Los Gatos, Podonos is building the infrastructure layer that decides whether voice AI feels real, human, and ready for primetime. The industry is racing toward a projected $47.5 billion by 2034, but growth without standards is chaos. Podonos is making sure the hype actually holds up when you press play. At the center is ✦Soohyun Bae, PhD in Computer Engineering from Georgia Institute of Technology and Y Combinator W22 alumnus. His track record stretches from engineering roles at Google Maps to leading AR mapping at Niantic, plus co-founding Bobidi and TickTock AI. He has seen what happens when technology scales without proper #guardrails. With Podonos, he's betting the future of AI isn't just about how smart the models get, but how believable and trustworthy they sound. Podonos evaluates models across naturalness, similarity, emotion, recognition accuracy, pronunciation, tone, and resilience. Add in #personaconsistency, and suddenly you're not just testing machines, you're testing their ability to pass as human. The twist? They deliver results in under twelve hours, when the legacy approach drags on for two months. In AI time, that's the difference between relevance and irrelevance. Investors caught the beat. Serac Ventures, led by Kevin Moore, took the lead, backed by NAVER D2SF, the venture arm of South Korea's NAVER, and KAIST Venture Investment Holdings. Together, they've now backed Podonos twice, with this $2.4 million round stacking on top of a $750,000 raise in March 2025. Total funding sits at $3.15 million, and every dollar is pointed toward scaling #engineering, refining #analysistools, and launching into the U.S., Europe, and Southeast Asia. This isn't theory. Podonos already serves Resemble AI, Play AI, and Sanas AI, hitting six-digit ARR by November 2024 with just four employees. Their secret weapon is reach: 150,000 evaluators spread across nine languages and thirteen locales, forming a human-in-the-loop platform that blends speed, rigor, and global coverage. That reach lets Podonos deliver evaluations at a scale and pace that no competitor has matched. The strategy is ambitious: expand into high-demand sectors like #healthcare, #finance, #gaming, and #advertising, then move beyond voice into #multimodal model evaluation, spanning video AI and large language models. The mission isn't just to check if AI works, it's to decide if AI works for people. #Startups #StartupFunding #EarlyStage #VentureCapital #PreSeed #AI #AIInfrastructure #Voice #VoiceAI #VoiceTech #Infrastructure #Technology #Innovation #TechEcosystem #StartupEcosystem #Hiring #TechHiring If software engineering peace of mind is what you crave, Vention is your zen.
1 Comment -
Preeti Cholleti
Nvidia Commits $5 Billion Investment in Intel to Boost AI and PC Product Collaboration https://lnkd.in/g2cMvfAT Nvidia and Intel’s $5 Billion Collaboration: A New Era for AI Innovation In an exciting turn for the tech industry, Nvidia and Intel have announced a groundbreaking partnership worth $5 billion, aimed at accelerating advancements in artificial intelligence. This collaboration focuses on harnessing cutting-edge technologies to enhance AI deployment across various sectors. Key Highlights: Strategic Investment: The partnership symbolizes a massive financial commitment to advance AI technologies. Joint Innovation: Together, Nvidia and Intel plan to push the limits of AI application in business, healthcare, and beyond. Boosting Competitiveness: This move strengthens the duo’s position in the rapidly evolving AI landscape, challenging other tech giants. AI enthusiasts and professionals should pay close attention to this dynamic alliance, as it promises significant impacts on future innovations. Explore the full implications of this partnership! Share your thoughts in the comments below and stay engaged as we witness this revolutionary journey in AI together! Source link https://lnkd.in/g2cMvfAT
-
Sinan AlRubaye
Years of progress, and we’ve only scratched the surface. Generative AI isn’t just changing workflows; it’s reshaping what’s possible. The future belongs to those willing to learn, adapt, and build with curiosity—not ego. #GenAI #Innovation #FutureOfCreation #nextgenpipeline #contentcreation #media #entertainment #film #tv #animation #vfx #games #ads #marketing
2 Comments -
Dustin Judge
AMD Ryzen Z1 Driver Support Problem Hurts ROG Ally Handheld PCs have quietly become the new frontier of gaming. They’re not going away anytime soon, and honestly, they shouldn’t — they’re fun, flexible, and finally powerful enough to matter. But if you own one powered by an AMD Ryzen Z1 or Z1 Extreme, like the Asus ROG Ally, you might want to check something before you boot up your next big release. Not a setting. Not a feature. Something far more important: driver support. Or rather, the complete lack of it. It wasn’t until Resident Evil Requiem launched that the cracks really started to show. The Ally wasn’t just struggling — it was collapsing. And suddenly, the rumor that AMD has quietly abandoned its Z1 APU went from “internet speculation” to “oh, this is actually happening.” Why Resident Evil Requiem Exposed AMD’s Z1 Problem When Requiem dropped, PC gamers immediately noticed something was off. The ROG Ally — a device that should be punching well above its weight — couldn’t even run the game at the lowest settings with upscaling enabled. TechPowerUp reported it. Social media confirmed it. But the truth is, the signs were already there. If you own a Z1 or Z1 Extreme handheld, you’ve probably noticed the same thing everyone else has: no GPU driver updates in over six months. Not a hotfix. Not a compatibility patch. Nothing. Pair that with Requiem’s disastrous performance, and suddenly the rumor that AMD has abandoned the Z1 doesn’t just make sense — it’s the only explanation left standing. The Performance Gap That Makes No Sense I test games on the ROG Ally constantly, so I’m used to its quirks. But Requiem was the first time the device felt like it was actively fighting the game. At 720p, lowest settings, upscaling on, the Ally struggled to break into double‑digit framerates. It felt broken. At first, I assumed the game was simply too demanding for a 2023 handheld. Then Wes — who was supposed to be relaxing in Hawaii — fired it up on his Steam Deck and casually pulled 40 fps. Let me be very clear: The Z1 Extreme is significantly more powerful than the Steam Deck’s APU. There is no universe where the Deck should outperform the Ally this dramatically. Unless the Deck is getting fresh, optimized drivers… and the Ally isn’t. Six Months Without Drivers — and It Shows This isn’t an Asus problem. The Lenovo Legion Go — another Z1 Extreme handheld — is stuck on the exact same outdated GPU driver from last September. Two different devices. Two different manufacturers. One common denominator: AMD’s Z1 APU. Meanwhile, Valve is updating the Steam Deck’s Mesa drivers constantly, rolling in new optimizations and fixes every few weeks. The Deck is aging gracefully. The Z1 handhelds are aging like milk. And the performance gap is widening with every new release. Valve Updates the Steam Deck, AMD Leaves the Z1 Behind Image of Asus ROG Ally, Courtesy Asus Here’s the part that makes this whole situation feel like a cosmic joke: The Z1...
Explore top content on LinkedIn
Find curated posts and insights for relevant topics all in one place.
View top content