Signal and Stories

Signal and Stories

By Boris Meinardus and Max BuckleyBusinessTechnologyCareers
Download on the App Store

Signal and Stories episodes

  • Principal AI Expert: The "Human In The Loop" Safety Mandate Is A Scam - Maria Sukhareva

    When tech giant narratives push stories about rogue AI agents hacking platforms and threatening human existence, it is easy to mistake wasteful brute-force search loops for genuine superintelligence. Behind the sensational headlines lies a much simpler technical reality: models trained for persistence will break rules, spam local systems, and hit every red light just to reach a benchmark goal.

    With over 15 years in natural language processing (spanning foundational academic research to serving as a Principal AI Expert at a major German enterprise), Maria Sukhareva launched The AI Realist to cut through industry hype. With her LinkedIn profile reaching over 40,000 followers, she helps executives, researchers, and developers separate true agentic utility from expensive, hype-driven circus acts.

    In this episode, we sit down with Maria to look at what is actually happening under the hood of modern AI agents, local hardware bottlenecks, and enterprise deployment failures.

    We break down a lot of fascinating tech and stories:
    ◼️ Brute Force vs. Intelligence: How training models for persistence causes them to ignore instructions, cheat benchmarks, and pollute local logs while appearing artificially smart.
    ◼️ Agentic Cyber Attacks: Why autonomous multi-agent loops will spark a wave of corporate data leaks, and why traditional "human-in-the-loop" governance fails as a security defense.
    ◼️ The Reality Of AI In Traditional Large Enterprises: How the largest and wealthiest German enterprises resist almost any kind of AI adoption. ◼️ The Local Hardware Fallacy: Why running local agent workflows on desktop setups or dedicated hardware quickly hits VRAM, context window, and bandwidth bottlenecks. Beyond the technical benchmarks and corporate balance sheets, this conversation serves as a grounded reminder that fluency does not equal intelligence or intent. Understanding AI as a specialized tool rather than an anthropomorphic entity is the only way to build software that actually delivers long-term value.

    Check out Maria Sukhareva's work:
    ◼️ The AI Realist: https://airealist.org/
    ◼️ LinkedIn: https://www.linkedin.com/in/msukhareva
    ◼️ Open-Weight AI Evening in Munich, 19 November 2026: https://airealist.org/events/taste-wain/
    ◼️ Agentic AI Masterclass in Zürich, 4 February 2027: https://airealist.org/workshops/agentic-ai-masterclass-zurich/

    Follow us:
    Max - LinkedIn: https://www.linkedin.com/in/maxbuckley/
    Max - Twitter/X: https://x.com/MaxWBuckley
    Boris - LinkedIn: https://www.linkedin.com/in/boris-meinardus-ba2302177/
    Boris - Twitter/X: https://x.com/BorisMeinardus 

    1 hr 30 min
  • Ex-AI Lead: I Got Fired, Signed a Book Deal & Am Launching a Startup in 42 Days! - Max Buckley

    When Max Buckley was abruptly fired from a $250M-funded AI startup while on holiday in Korea, he didn't just lose his job. He watched an entire remote office equipped with top-tier hardware get shut down overnight without warning.

    Instead of slowing down, the former Google engineer turned his termination into a six-week whirlwind, securing a book deal with O'Reilly, bootstrapping an edge-AI robotics startup, and receiving over 600 VC pitches the second he flipped his status to "stealth".

    In this episode, we sit down with Max to look at what actually happens inside hyper-funded AI startups, how to bounce back from sudden corporate layoffs, and why the European startup ecosystem might be uniquely positioned to lead the next wave of real-world AI.

    We break down a lot of fascinating tech and stories:
    ◼️ Abrupt Startup Restructuring How EXA raised a quarter-billion dollars and kitted out a remote office only to about ship four months later.
    ◼️ The Surge of Venture Capitalists What actually happens when you change your status to "stealth startup" and get spammed by 600 VCs overnight.
    ◼️ European Startup Landscape Why Zurich's 250-company robotics cluster gives Europe a distinct competitive edge over Silicon Valley in physical AI and more.
    ◼️ O'Reilly Publishing Blueprint How turning a rejected book proposal into a proofreading gig led to a signed contract with technical publishing royalty.

    Beyond the startup chaos, Max's journey is a masterclass in career resilience—proving that building public credibility and staying adaptable will always outpace corporate volatility.

    Follow us:

    Max - LinkedIn: https://www.linkedin.com/in/maxbuckley/

    Max - Twitter/X: https://x.com/MaxWBuckley

    Boris - LinkedIn: https://www.linkedin.com/in/boris-meinardus-ba2302177/

    Boris - Twitter/X: https://x.com/BorisMeinardus

    59 min
  • Cognition AI Engineer: Inside The $26B AI Tool Replacing Enterprise IT – Ben Lau
    When Ben Lau rose to Junior Partner after a decade at McKinsey, he realized a fundamental tension in corporate consulting: selling high-priced strategic slide decks was becoming increasingly detached from hands-on technical execution.
    Instead of remaining in comfortable upper-tier management, he made a radical choice to reset his career—taking on a coding challenge that landed him as a Forward Deployed Engineer at Cognition, the $26B startup behind Devin.
    In this episode, we sit down with Ben to look at the seismic shift transforming software engineering and enterprise strategy:
    We break down a lot of fascinating tech and stories:
    ◼️ McKinsey To AI Startup: Why he walked away from a 10-year career as a Junior Partner to jump directly into the high-stakes world of frontier AI deployment.
    ◼️ The Rise Of The FDE: How Forward Deployed Engineers are replacing traditional management consulting by delivering working software instead of expensive slide decks.
    ◼️ Cloud Agents Vs. Local Tools: Why local coding assistants fall short on massive enterprise repositories, and how cloud-hosted agents leverage elastic virtual machines to handle complex parallel workloads.
    ◼️ Eliminating AI Slop: The exact architecture Cognition uses to keep agents from duplicating code, from pre-indexing codebases with DeepWiki to enforcing multi-agent dialectic code reviews.
    ◼️ Non-Engineers Writing Code: How non-technical sales and operations teams are using Devin to diagnose system bugs, run tests, and submit functional pull requests directly.
    Beyond the technical discussion, this chat is a in inspiration when it comes to shifting careers! It shows that you don’t have to be a deeply technical person to be able to work with code anymore! It is a story about finding out what is important to each person, how to provide value, and how to adapt to modern tools to solve those problems.
    Enjoy!
    Check out Cognition and Devin at cognition.com and devin.ai
    1 hr 17 min
  • ML Systems CEO: NVIDIA's Monopoly Will Have a Problem - Emilio Andere
    NVIDIA currently charges the entire tech industry a massive 60%* profit margin on AI chips. But behind closed doors, their software monopoly is starting to slip as new AI agents make it easy to run alternative hardware.
    Growing up in Mexico, Emilio Andere realized early on that local opportunities in deep tech were heavily limited, prompting a move to the US to study mathematics and publish ML security research at NeurIPS. Now, he is the co-founder and CEO of Wafer, a YC and Jeff Dean-backed startup automating low-level kernel engineering to actively erode the CUDA moat and build a future of open, heterogeneous AI hardware.
    In this episode, we sit down with Emilio Andere to look at the future of heterogeneous hardware, the brutal reality of working with Y Combinator, and how they went from a simple idea to serving billions of tokens a day on Open Router.
    We break down a lot of fascinating tech and stories:
    ◼️ The CUDA Moat - Why NVIDIA's margins are actually a soft software moat that is easier to break than the market realizes.
    ◼️ Kernel Engineering - What it is, the state of AI-automated design, and the challenges of designing Kernels on hardware other than NVIDIA.
    ◼️ Y Combinator - The brutal reality of internal competition amongst startups that forces founders to grow rapidly or face public failure.
    ◼️ The Open Router Surge - What happened behind the scenes when Wafer went live on the world’s biggest token router and instantly processed billions of tokens.
    Beyond the raw software and hardware deep-dives, this conversation is about the democratization of human intelligence. By breaking down this huge monopoly, we are opening up a world where advanced AI is cheap, accessible, and sovereign for everyone on Earth.
    Follow Emilio Andere's work:
    ◼️ LinkedIn: https://www.linkedin.com/in/emilio-andere/
    ◼️ Wafer: https://www.wafer.ai/
    Further reading:
    ◼️ Kernel Bench paper: https://arxiv.org/abs/2502.10517
    ◼️ *https://newsletter.semianalysis.com/p/ai-value-capture-the-shift-to-model
    ◼️ https://www.wafer.ai/blog
    1 hr 17 min
  • AI Expert Warning: The Permanent Underclass Discourse Is A Lie, Do This Instead!
    When Devansh Devansh was part of a small team that beat Apple at detecting Parkinson’s disease using real-time voice calls, he learned a frustrating lesson about big tech: internal corporate politics can easily buy up a breakthrough algorithm and bury it.
    Years later, after building a massive following on his Substack, AI Made Simple, he made a radical choice. Instead of chasing a cozy role at a major AI lab, he chose to open-source his research and build a decentralized community called the Chocolate Milk Cult. His reasoning? Big tech companies have become too risk-averse, with researchers more focused on job security than making actual breakthroughs.
    In this episode, we sit down with Devansh to cut through the hype and look at the real engineering bottlenecks in AI today.
    We talk about:
    ◼️ What actually happens when corporate bureaucracy sidelines life-saving technology.
    ◼️ The practical flaws behind how OpenAI and Anthropic handle AI memory, and how his team solved it for legal tech.
    ◼️ Why he walked away from institutional big tech money to keep his research open to everyone.
    ◼️ How big tech's fear of making mistakes is quietly stalling the next generation of AI models.
    ◼️ The current "underclass" AI narrative, and what it really takes for a young engineer to stand out right now.
    If you're tired of the generic AI hype cycle and want a realistic look at where the technology is actually going, this conversation is for you.
    Follow Devansh's Substack:
    https://www.artificialintelligencemadesimple.com/
    1 hr 32 min
  • Behind the Scenes: Working At Top AI Startups
    What is it like to work at a cutting-edge AI startup?
    AI Researcher BORIS MEINARDUS and Google-engineer-turned-startup-Head-of-AI MAX BUCKLEY go behind the scenes of the AI industry — from startup towers in the valley to packed poster sessions in Rio, and what it actually takes to stand out when there are 400+ applicants for two open roles.
    Max Buckley, a Senior Machine Learning Engineer at Google with over 12 years of experience, has now transitioned to leading the Zurich office of Exa, a multi-million-dollar, YC-backed AI startup.
    He shares behind-the-scenes insights into what it is like to lead such a startup, which involves traveling around the globe and interviewing hundreds of applicants.
    Boris has also been traveling around the globe, from Tokyo to San Francisco to Rio and back.
    We discuss:
    ◼️ What it's really like to move from Google's AI culture to a cutting-edge AI startup
    ◼️ The hiring reality: 400+ applicants, two open roles, and the detail that costs you the offer
    ◼️ Why poster sessions at top conferences are half-empty — and how Boris drew crowds at ICLR
    ◼️ Boris's paper: how an ensemble of 8 small models can match GPT-4o across diverse tasks
    ◼️ Max's 5 open-source PRs to Microsoft — and why fixing one app fixes them all
    1 hr 26 min
  • My Final Day at Google After 12 Years (Senior ML Engineer)
    Google Senior ML Engineer MAX BUCKLEY and AI Researcher BORIS MEINARDUS expose why being a "top contributor" on your team won't save your career, how to break free from the "hand-holding" trap of big tech, and the high-agency playbook for taking total ownership of your professional trajectory in the age of AI.
    Max Buckley is a Senior Machine Learning Engineer at Google with over 12 years of experience who has now transitioned to leading the Zurich office of a multi-million-dollar, YC-backed AI startup.
    Having navigated 10 different roles within the tech giant, Max has developed a reputation for identifying the "signals" of career success that go far beyond simple coding proficiency.
    We explain:
    ◼️ Why your manager is the "Source of Truth" (and how that can be used against you)
    ◼️ Big Tech vs. Startups: The hidden psychological cost of "structured" environments
    ◼️ Why "hand-holding" is the silent killer of senior engineering growth
    ◼️ The Ownership Framework: How to move from being a "hired hand" to a project visionary
    ◼️ Why taking your career into your own hands is the only way to survive the AI transition
    Foundations of Vector Retrieval - Sebastian Bruch
    https://arxiv.org/abs/2401.09350
    1 hr 33 min
  • The Man Automating Coding: Lessons Learned Building a SOTA Coding Agent (#2 - Nicolay Gerold)
    AMP founding engineer NIKOLAY GEROLD reveals the "unsolved problem" of autonomous code deployments, why AMP is the PORSCHE OF CODING AGENTS, and the high-risk future of open-source personal assistants with unrestricted access to your entire life.
    Nikolay Gerold is a German developer and the Founding Engineer at Amp, a frontier AI startup building a high-performance coding agent that competes with popular tools like Claude Code. With years of experience in applied AI, he is currently pioneering the infrastructure that allows agents to move beyond "vibe coding" and into autonomous system architecture.
    He explains:
    ◼️ Why Amp is the "Porsche" of coding agents while others risk becoming "Homer’s car."
    ◼️ The Unsolved Problem: When is it safe for an agent to push to production vs. using a Pull Request?
    ◼️ The "OpenClaw" Experiment: Extrapolating a future where AI has access to your entire life and data.
    ◼️ Why the most powerful (and risky) agents will likely come from open-source, not Big Tech.
    ◼️ The "EU Inc" reality: How to build and scale AI startups within European regulations.
    ◼️ Why developers won't be replaced—if they know how to navigate the "agentic" shift.
    Check out Nicolay's Podcast: @howaiisbuilt
    Follow us on our other socials:
    Max - LinkedIn: https://www.linkedin.com/in/maxbuckley/
    Max - Twitter/X: https://x.com/MaxWBuckley
    Boris - LinkedIn: https://www.linkedin.com/in/boris-meinardus-ba2302177/
    Boris - Twitter/X: https://x.com/BorisMeinardus
    1 hr 31 min

About Signal and Stories

From the publisher's feed

The signal behind the AI hype and the stories of the people building it.