AI & Robotics

About AI & Robotics

Robotics, AI, and computer vision, explained for builders, researchers, and the technically curious.

Publisher

Asad Norouzi

Category

technology

Language

en

Explicit

No

First episode

16 Jul 2026

Latest episode

31 Oct 2026

Latest episodes

7 episodes in the feed.

1. 

   31 Oct 2026

   The Death of LeetCode: How to Actually Hire AI & ML Engineers (Best for reach & curiosity)

   If an AI tool can solve your hardest 60-minute LeetCode challenge in three seconds and generate an ATS-optimized resume in five, what are your hiring filters actually measuring? In this deep dive, we break down why traditional dot-com-era tech hiring playbooks fail catastrophically when evaluating machine learning and artificial intelligence talent. Relying on syntax memorization and whiteboard puzzles to hire ML engineers is like using a spelling bee to hire a novelist! It tests the wrong skillset while filtering out the practitioners who actually know how to build resilient, real-world systems. We unpack an evidence-based blueprint for identifying, vetting, and retaining top-tier AI engineers: The Fall of the Old Filters: Why AI-polished resumes and algorithmic coding puzzles fail to measure non-deterministic failure modes, data drift, and production ML realities. The 3 True Predictors of ML Success: How to evaluate curiosity with follow-through, communication precision across non-technical stakeholders, and productive uncertainty tolerance. The 4-Round Modern Interview Loop: Moving away from multi-day whiteboard hazing toward real backlog problems, system design at scale, and adversarial collaboration simulations. The Pedigree Trap: Navigating the stark operational divide between academic research novelty and production engineering constraints (cost, latency, and reliability). Ditching "Culture Fit" for "Culture Add": Replacing vague affinity bias with structured, objective behavioral rubrics that protect team diversity and psychological safety. The 30-60-90 Day Retention Architecture: Why onboarding is an engineering problem—and how small, early production wins beat imposter syndrome and stop first-year turnover

2. 

   29 Sept 2026

   How Self-Driving Cars Actually See the World: Inside the Perception Stack

   When we imagine an autonomous vehicle, it is common to unconsciously assume the car simply "opens its eyes" to interpret the road. The engineering reality requires pushing the boundaries of physics, optics, and artificial intelligence to build a reliable, millimeter-accurate digital representation of a chaotic physical environment. This episode demystifies the mechanics of autonomous vehicle perception by breaking down the complex hardware and algorithmic frameworks that allow machines to physically construct their reality and safely navigate urban environments. Episode highlights: The Hardware Trinity: A breakdown of passive cameras capturing high-resolution spatial data, LiDAR generating absolute 3D geometric point clouds via time-of-flight laser pulses, and Radar cutting through severe weather to provide instantaneous Doppler velocity. Sensor Fusion: The mathematical heavy lifting required to map disparate 2D pixels and 3D point clouds into a unified top-down Bird's Eye View (BEV). Object Permanence in AI: How spatiotemporal transformers and recurrent memory states act as the vehicle's short-term memory, allowing it to continuously track a vehicle even when it is temporarily occluded by a massive semi-truck. Vehicle-to-Everything (V2X): The next frontier of autonomous engineering, where cars will wirelessly share their fused perception data to eliminate blind spots and effectively see through solid infrastructure.

3. 

   19 Aug 2026

   Brain Meets Body: Unpacking the Pi0 Robotics Revolution

   For years, Artificial Intelligence has been the ultimate "digital brain", mastering everything from writing poetry to generating hyper-realistic videos. But what happens when you give that genius-level brain a physical body? In this deep dive, we break down the groundbreaking Pi0 paper from Physical Intelligence. Discover how researchers are bridging the massive gap between digital reasoning and physical dexterity, finally allowing AI to step out of the internet and into our messy, unpredictable physical world. We’re talking less "clunky robot dropping your coffee" and more "fluid, 50Hz continuous motion." In this episode, we unpack: The Brain & Cerebellum Approach: How combining massive Vision-Language Models (VLMs) with a specialized "Action Expert" is changing the robotics game. Flow Matching Explained: Why standard text-generation math is completely wrong for physical movement, and what researchers are using instead. 10,000 Hours of Trial & Error: Inside the massive "OXE Magic Soup" dataset and the sandbox pre-training that teaches robots about gravity the hard way. The Future of Movement: Could an AI with 10,000 lifetimes of physical experience invent entirely new ways to manipulate objects that human biomechanics never even considered? Whether you're an AI researcher, a tech enthusiast, or just someone eagerly waiting for a robot to do your chores, this episode explores the bleeding edge of embodied AI.

4. 

   12 Aug 2026

   From Cruise Ships to Hot Air Balloons: The AI Manager's Playbook

   Are you trying to manage your AI team like a traditional software engineering crew? If so, your project is already sinking. Transitioning from managing traditional full-stack developers to leading AI and machine learning engineers requires a complete paradigm shift. It is the difference between captaining a highly predictable, mechanical cruise ship and trying to steer a hot air balloon by reading the wind. In this deep dive, we unpack the hidden traps that catch experienced engineering managers off guard when they step into the probabilistic world of AI. We break down the evolution from Software 1.0 to Software 3.0 and provide a tactical, hands-on operational playbook for leading data scientists, researchers, and engineers without bankrupting your project or your sanity. In this episode, we cover: The Tech-Lead Trap: Why diving into the code to "unblock" your team fails miserably against a wall of opaque model weights and statistical probabilities. Monte Carlo Simulations & Trust Budgets: How to handle project estimations and stakeholder updates when standard Agile sprints and hard deadlines are no longer viable. Researchers vs. Engineers: The root cause of the most common organizational friction in AI and how to establish clear "swim lanes" and strict handoff APIs. Psychological Safety as a Technical Tool: Why building a culture that treats failures as data, not personal attacks, is critical to catching catastrophic system errors before production. The Future of Management: What happens when your direct reports are no longer human, but autonomous AI agents?

5. 

   31 Jul 2026

   Code vs. Concrete: Why Your AI Can't Fold a Towel

   AI can write a flawless Shakespearean sonnet in two seconds, but ask it to fold a laundry towel and the entire system violently crashes. Why? In this episode, we cut through the Silicon Valley hype machine and dive into the brutal math of the physical world. While the tech industry pours billions into chatbots and generative software, we explore the massive 80% failure rate of enterprise AI and the terrifying reality of the "Sim-to-Real Gap." Join us as we break down why the true future of artificial general intelligence (AGI) won't emerge from a disembodied server farm, but from robots learning how to fail, fall, and feel gravity. If you want to know what happens when code actually hits concrete, this is the ultimate reality check. In this episode, we unpack: The Sim-to-Real Gap: Why robotic brains that ace digital simulations instantly crash when faced with real-world physics and unexpected friction. Vision-Language-Action (VLA) Models: How mapping pixels and text to actual physical motor torque is revolutionizing machine learning. The Hardware Bottleneck: Why the smartest software in the world is useless if it shatters a titanium arm on a desk, and how bio-hybrid designs and optical tactile sensors (like GelSight) are saving the day. Embodied Cognition Theory: The philosophical shift proving that true intelligence requires physical consequence. Listen now to discover why the most profound revolution in technology isn't happening behind a screen. It's learning how to pick up a box. (Don't forget to follow, rate 5 stars, and share this episode with the tech optimist in your life!)

6. 

   22 Jul 2026

   Embodied: The Brutal Physics of Robotic AI

   Digital AI is plateauing, but physical robotics is just having its "ChatGPT moment." Why is the tech industry suddenly desperate to stuff fragile, hallucinating software into 200-pound heavy metal bodies? In this deep dive, we cut through the sci-fi hype to uncover the brutal reality of "Embodied AI." We explore why teaching an AI to walk is a mathematical nightmare and how the physical world acts as the ultimate, unforgiving fact-checker for language models. From the engineering nightmare of the "mass penalty spiral" to the explosive dangers of the "reflected inertia trap," we break down the exact physics keeping robots from taking over your kitchen for now. In this episode, we cover: The Software Failure Rate: Why 80% of corporate AI projects are failing and how physical embodiment is the proposed cure. The Walking Problem: The brutal math of neural scaling laws and why it takes 24 times the data to make a robot move. The $16,000 Humanoid: A look at the sudden democratization of hardware (like the Unitree G1) and how university labs are replacing massive corporate budgets. Emergent Capabilities: The haunting reality of what happens when a physical machine suddenly figures out a complex physical task entirely on its own. Grab your headphones and join us as we explore the exact intersection where digital brains crash headfirst into the laws of physics. #ArtificialIntelligence #Robotics #EmbodiedAI #TechNews #Engineering #FutureOfWork #MachineLearning

7. 

   16 Jul 2026

   EmbodiedAI

   A comprehensive look at the artificial intelligence and robotics landscape in 2026, focusing on industry predictions, financial valuations, and emerging technological trends. One report details twelve specific forecasts for embodied AI, highlighting the rise of Vision-Language-Action models, the push for robotics-specific hardware, and the persistent challenges of safety and long-horizon task execution. A secondary analysis warns of a potential market consolidation, where redundant development may cease as investors prioritize startups with clear commercial delivery and healthy cash flows. Together, these illustrate a shift from experimental research to a high-stakes industrial era defined by autonomous agents, specialized vertical platforms, and a global race for AI supremacy.

Contact AI & Robotics

Guest appearances

Does not typically book guests

Based on episode analysis; this does not confirm that the show is currently accepting guests.