Coredump Sessions

Summary:In today's Coredump Session, we sit down with Noah Pasek-Nelson and Chris Markus, the co-founders of BootLoop: an AI-powered platform purpose-built for firmware teams. The conversation gets into the real mechanics of how AI actually fits into embedded development workflows, from hardware bring-up and driver generation to hardware-in-the-loop testing and field debugging. Noah and Chris bring hard-won perspective from SpaceX, MIT, and FDA-regulated medical devices, and they don't shy away from the messy questions— hallucination, autonomy, trust, and what it actually takes to close the loop when your runtime is physical hardware.Key Takeaways:BootLoop is an AI platform built for the entire firmware lifecycle, from hardware bring-up through field debugging, not just code generation.The two core pillars are hardware understanding (ingesting schematics, data sheets, netlists) and hardware interaction (oscilloscopes, logic analyzers, GDB).Firmware engineers only spend about 20% of their time writing code. The rest is testing, debugging, and validation, which is where BootLoop focuses.Agents do their best work in tight feedback loops. BootLoop builds, flashes, and tests on device automatically so the agent can iterate without human intervention.Hardware understanding eliminates hallucination by grounding the model in your specific chips, register maps, and pin configurations rather than public code averages.Power optimization is a standout use case: feed in the data sheet, connect a power meter, set a target, and come back to optimized code.BootLoop supports C, C++, Rust, Zephyr, bare metal, and embedded Linux. HDL and FPGA support is coming.Sentinel, their field debugging product, automatically kicks off root cause analysis the moment a bug comes in from the field, before an engineer even sits down.Pricing is a flat fee with unlimited usage, intentionally aligned so the incentive is correct code, not more code.New chips are supported as fast as you can upload the documentation. There is no supported chips list.Chapters:00:00 Intro: AI at the Bench for Firmware Teams02:23 What Is BootLoop? Meet the Founders05:18 What Actually Changes Day-to-Day for Firmware Engineers07:13 The Hallucination Problem and How to Solve It in Hardware09:51 From Data Sheet to Flashed Device: A Live Workflow Demo11:20 Oscilloscopes, GDB, and Natural Language: Hardware Interaction Explained13:34 How Much Babysitting Does an AI Agent Actually Need?45:59 Greenfield vs. Brownfield: Where BootLoop Fits Best47:33 Audience Q&A: Business Model, Security, and Supported Chips57:35 Debugging in the Field: Sentinel, MCP, and Root Cause Analysis1:00:04 Closing Thoughts and How to Get a DemoJoin the Interrupt Slack ⁠⁠⁠⁠Watch this episode on YouTube [ADD Playlist YT Link from here)⁠⁠Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

SummaryIn today’s Coredump Session, François and Chris wrap up the year with a special milestone celebration for the 20th episode of the series. They take listeners behind the mic to reflect on how Coredump began, the moments that defined 2025, and the lessons learned along the way. Expect highlights from the year’s most talked-about discussions, bold predictions for what’s ahead in 2026, and the biggest live Q&A yet—where nothing is off the table.Key Takeaways:Why Coredump Sessions started and how the show has evolved over 20 episodesThe biggest engineering lessons guests shared throughout 2025The most surprising device failures and field behaviors teams faced this yearHow firmware teams adapted to fast-changing toolchains and new requirementsThe rise of AI-assisted debugging, testing, and development in embedded workHow security mandates tightened and reshaped device developmentThe recurring theme that real-world conditions rarely match lab assumptionsWhat François and Chris learned from producing the series behind the scenesWhat the hosts expect to define embedded systems development in 2026The expanded live Q&A format and how the community is shaping future episodesChapters:00:00 Teasers01:30 Intro/ Welcome03:35 The Genesis of Core Dump06:54 Favorite Moments and Lessons Learned08:45 Clip Reaction: Nick Sinas's OTA Nightmare in the Snow12:19 Clip Reaction: Lack of Security at Pebble16:45 Clip Reaction: Dan Mangum on AI & Engineering21:50 Clip Reaction: Vatsal at Ultrahuman Talks Always-On Devices25:29 Clip Reaction: Chad from Gabb Talks Security30:12 Looking Ahead: Predictions for 202635:59 The Rise of AI and Edge Computing42:24 Evolving Skills for Firmware Engineers47:01 Security in Embedded Systems49:01 The Future of Testing in Embedded Development53:15 Conclusions & Thank Yous⁠Join the Interrupt Slack⁠ ⁠⁠⁠⁠⁠⁠⁠Watch this episode on YouTube ⁠Follow Memfault⁠⁠⁠LinkedIn⁠⁠⁠⁠⁠⁠Bluesky⁠⁠⁠⁠⁠⁠Twitter⁠⁠⁠Other ways to listen:⁠⁠⁠⁠⁠Apple Podcasts⁠⁠iHeartRadio⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Amazon Music⁠⁠GoodPods⁠⁠Castbox⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Visit our website

In today's Coredump Session, François Baldassari and Chris Coleman sit down with Ultrahuman co-founder Vatsal Singhal to unpack what it takes to build and scale a hardware startup in the fiercely competitive health wearable market. From transitioning from software to hardware to building responsibly with AI and machine learning, Vatsal shares what it means to blend deep engineering rigor with a mission to improve human performance. This conversation explores the challenges, surprises, and future of health-tech innovation at the edge.Key TakeawaysHow Ultrahuman transitioned from a software-first mindset to mastering complex hardware development.Lessons learned moving from large-scale software systems to building precision-focused health wearables.Why building hardware for health requires a fundamentally different level of accountability and rigor.The role of machine learning at the edge and how it enables better, faster insights while managing battery and compute tradeoffs.How responsible use of AI in health applications shapes product design and user trust.The importance of rapid iteration cycles and adopting software methodologies in hardware innovation.Insights into how Ultrahuman’s internal teams use AI not just in engineering, but across all business functions.A look at what’s next for health-tech — and where innovation is heading in wearables and bio-sensing.Chapters:00:00 Intro & Teasers03:43 From Software to Hardware: The Leap of Faith07:49 The Harsh Realities of Hardware10:40 Iterating Fast Without Breaking People15:17 Redefining A/B Testing in Hardware21:40 Why Ultrahuman Built Its Own Factory26:56 Scaling Production Across Continents29:48 Managing Complexity: 20 Hardware Revisions in a Year35:08 Firmware Velocity & Observability with Memfault43:42 Health Tech Meets Regulation47:55 Shared Codebases & Fast Iteration Across Products50:39 Building the Machines That Build the Rings54:34 Responsible AI & The Future of Health Wearables56:35 Closing Reflections & Key TakeawaysJoin the Interrupt Slack ⁠⁠⁠⁠Watch this episode on YouTube Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

In today’s Coredump Session, we explore the rise of kids safe tech with leaders from the GABB team, creators of connected devices designed specifically for children. From designing products that prioritize child safety to integrating AI in ways that support families, this conversation unpacks the complexities of building secure, intuitive technology for the next generation. The team also shares real-world lessons on hardware partnerships, customer trust, and what it takes to innovate responsibly in the IoT space.Key Takeaways:Kids safe tech is an emerging category centered on digital safety for children.IoT connectivity is essential for building trust between parents and kids.Gabb’s mission is rooted in protecting families through intentional technology.Strong partnerships are critical to scaling hardware, firmware, and software development.AI enhances product safety by filtering content and flagging risks in real time.Security and data privacy are foundational to Gabb’s product design philosophy.Product development requires aligning launch timing, market needs, and platform strategy.Managing multiple product lines demands balance between innovation and sustainment.Kids often outpace parents in tech fluency—celebrating young users can inspire product direction.Efficient QA processes are necessary to uphold product quality and customer trust.Customer feedback and word of mouth are vital inputs for roadmap decisions.Data minimization is a core principle when designing for young users.Cross-functional collaboration drives more effective and family-first product development.Chapters:00:00 Episode Teasers & Intro04:46 Understanding Kids Safe Tech10:25 The Role of Partnerships in Product Development14:59 Navigating AI in Product Design20:20 Balancing Needs of Kids and Parents28:14 Ensuring Security in Kids Tech32:31 Celebrating Advocacy and Security Solutions33:45 Navigating Privacy in Child Analytics37:30 Product Development Cycle and Timelines41:31 Balancing Current and Future Product Development45:53 Sustaining Products Amid New Launches48:37 Customer-Centric Approach in Product Maintenance52:42 Firmware Versioning Challenges and StrategiesJoin the Interrupt Slack ⁠⁠⁠⁠Watch this episode on YouTube ⁠Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠

In today's Coredump Session, we dive into the fascinating world of building Edge AI models that truly work in real-world environments. Joined by David Tischler, Developer Program Manager, and Alessandro Grande, Head of Product at Edge Impulse (A Qualcomm Company), we unpack what it takes to deploy AI on tiny devices, explore practical applications from wearables to industrial use cases, and discuss why customization, hardware choices, and continuous monitoring are critical for success. Tune in to explore how Edge AI is transforming device development and enabling smarter solutions.Key Takeaways:Edge AI empowers devices to process data locally, significantly reducing latency, bandwidth usage, and improving privacy.The best use cases for Edge AI today often involve video and audio analytics, wearables, and industrial sensor applications.Customization is the key value of AI, making it easy to fine-tune models for specific tasks or customer needs without extensive traditional coding.Effective Edge AI requires thoughtful pre-processing (DSP), not just AI models—this combination significantly improves model performance.Hardware selection is crucial; developers must balance model complexity with device constraints, such as available RAM and compute power.Many AI co-processors marketed for embedded systems today are essentially DSP units rebranded as AI accelerators, and usability matters more than raw performance.Observability and OTA (over-the-air) updates are critical components in Edge AI deployment, enabling continuous monitoring, data-driven refinements, and quick responses to issues in the field.Production readiness in Edge AI involves not only initial deployment but ongoing data collection, model retraining, and continuous improvement cycles.Chapters:00:00 Intro & Teasers: Edge AI's Real-World Promise01:57 Meet Our Guests: David Tischler & Alessandro Grande from Edge Impulse05:19 How Edge AI Took Off: From Hyped to Essential09:21 Beyond Voice Commands: Emerging AI Use Cases12:02 Defining the Edge: Wearables to Factories19:09 AI's Hidden Superpower: Customization and Fine-Tuning26:15 Why AI Belongs at the Edge: Latency, Privacy, and Power28:38 Building the Software Stack: Edge AI for Embedded Engineers34:17 Choosing Your Hardware: Constraints and AI Accelerators45:42 Observability and OTA Updates: Essential for Edge AI52:28 Audience Q&A: Fine-Tuning, TinyML, and the FutureJoin the Interrupt Slack ⁠⁠⁠⁠Watch this episode on YouTubeFollow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

In today's Coredump Session, François and Chris share a major milestone in Memfault’s journey: Memfault is joining forces with Nordic Semiconductor. They’re joined by Kjetil Holstad from Nordic to explore Nordic’s history, how the acquisition came to be, and the exciting future they envision together. Along the way, they dive into the evolution of Bluetooth, the challenges of building complex embedded systems, and why developer experience is at the heart of it all.Key Takeaways:Nordic Semiconductor’s journey from design services to global Bluetooth leadership hinged on bold bets and developer-centric decisions.The inclusion of Bluetooth Low Energy in the iPhone was a pivotal moment that accelerated Nordic’s growth.Building robust embedded systems today requires deep attention to software complexity, which has significantly increased over time.Decisions that prioritize developer experience—like open documentation and accessible SDKs—have been instrumental to Nordic’s success.Memfault’s partnership with Nordic was driven by a shared vision to make embedded observability easy, accessible, and deeply integrated.Memfault’s pre-integration with Nordic’s SDKs and their collaborative engineering work have helped reduce barriers for developers.The vision for a “chip-to-cloud” platform is about delivering seamless device management, monitoring, and OTA updates with a scalable, unified solution.Both Memfault and Nordic are committed to supporting non-Nordic hardware, focusing on building solutions that serve the broader embedded ecosystem.Chapters:00:00 Episode Teasers & Intro02:59 The Origin Story of Nordic Semiconductor06:11 The Evolution of Bluetooth Technology at Nordic09:40 Bold Decisions and Learning from the Market14:24 Nordic's Commitment to Developers and Software17:32 The Shift Towards Software in IoT20:04 Embracing Complexity: The Future of Nordic's SDKs20:55 Understanding Trust Zone and Embedded Systems21:46 The Importance of Strategic Partnerships22:30 Building Relationships in the Tech Industry25:09 The Value of Collaboration and Integration27:38 Enhancing Developer Experience through Integration32:24 Announcing the Partnership and Future Vision33:07 Creating a Chip to Cloud Platform36:05 Supporting Non-Nordic Devices and Ecosystem Expansion39:10 Reactions to the Acquisition Announcement44:58 Q&AJoin the Interrupt Slack ⁠⁠⁠⁠Watch this episode on YouTubeFollow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

In today’s Coredump Session, we dive into the origins and evolution of Zephyr RTOS with Kate Stewart, VP of Dependable Embedded Systems at the Linux Foundation. From Intel’s early ambitions to a thriving global community, Kate unpacks how Zephyr grew into a leading open-source RTOS and what makes it uniquely resilient and developer-friendly. This conversation also explores the technical shifts shaping embedded development and how governance, safety, and collaboration continue to steer Zephyr’s trajectory.Speakers:Kate Stewart: Vice President of Dependable Embedded Systems, The Linux FoundationFrançois Baldassari: CEO & Founder, MemfaultThomas Sarlandie: Field CTO, MemfaultKey Takeaways:Zephyr was born from Intel’s desire for a scalable, secure, and open RTOS, evolving from Wind River roots.Early adoption of Linux-inspired practices, like Kconfig and "signed-off-by" contributions, lowered friction and encouraged community participation.The project’s governance model, emphasizing multi-vendor participation and elected leadership, prevents corporate capture and boosts resilience.Zephyr’s pragmatic reuse of tools like MCUboot accelerated development and expanded capabilities.Long-term support (LTS) releases—now extended to five years—make Zephyr production-friendly and aligned with regulatory demands like the CRA.Innovations like the Twister test framework and open testing infrastructure set Zephyr apart for visibility and maintainability.Zephyr thrives as complexity in embedded systems increases, filling the gap left by simpler RTOSes ill-suited for modern MCU workloads.Not every project is a fit for Zephyr—especially ultra-low-end 8-bit systems—but it excels in growing, connected device classes.Chapters:00:00 Introduction and Guest Introduction04:12 Building Zephyr: Intel’s Open RTOS Bet06:39 Governance That Guards Against Capture08:10 Borrowing From Linux, Avoiding Its Baggage09:41 What Makes Zephyr Different13:55 Zephyr in Production: LTS and Real-World Adoption16:15 Scaling with Twister and QEMU18:15 Taming Complexity Without Losing Performance35:45 SBOMs and the Future of Compliance38:20 A Head Start on Security Standards43:02 Inside Zephyr's Safety Certification Journey46:44 Real-World Use Cases and Industry Uptake50:25 What's Next for Zephyr and the RTOS Landscape53:12 Final Reflections and Closing Thoughts⁠⁠Join the Interrupt SlackWatch this episode on YouTube ⁠Suggest a Guest⁠Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

In today’s Coredump Session, we zoom in on the rapidly evolving world of cellular IoT—what’s working, what’s changing, and what developers should know. With expert insight from Fabien Korheim of ONES, the conversation breaks down MVNOs vs MNOs, dives into certification hurdles, explores connectivity trade-offs like NB-IoT vs LTE-M, and unpacks why cellular is quietly powering more devices than you think. Whether you're building metering devices or baby monitors, this one hits the full stack—from tech to business models.Key Takeaways:MVNOs simplify global IoT deployments by abstracting regional carrier relationships and reducing SKU complexity.LTE-M is currently the safest bet for low-power cellular applications, with 5G RedCap positioned as a future alternative.Certification processes are lighter with MVNOs, especially when using pre-approved modules.Cellular IoT is ideal where Wi-Fi isn’t guaranteed, like basements, forests, and mobile tracking.Consumer IoT has huge untapped potential—cellular can dramatically improve usability and reduce returns.Battery life and data costs are major design considerations, especially when scaling fleets globally.Multiradio devices and smart fallback strategies (e.g. BLE/Wi-Fi + Cellular) are becoming more common.Debugging tools and observability platforms are essential for maintaining reliability across networks, devices, and regions.Chapters:00:00 Episode Teasers & Intro02:34 MVNO vs MNO: What’s the Difference?06:28 Certifications, SIMs & Simplifying Deployment12:31 NB-IoT, LTE-M, LoRaWAN & Satellite—Explained23:43 5G for IoT: Hype or Here?27:14 Top Use Cases: Meters, Trackers & Wildlife33:28 The Big Opportunity: Cellular in Consumer Devices36:33 Business Models: Who Pays for Cellular?37:49 Getting Started: Kits, SIMs & Copy-Paste Firmware41:59 Common Mistakes & What to Watch in the Field47:15 What to Measure: Observability That Scales49:13 Q&A: Prioritization, Firmware Updates, RedCap & More⁠⁠Join the Interrupt SlackWatch this episode on YouTube ⁠Suggest a Guest⁠Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

REGISTER FOR PART 2 OF THE PEBBLE CONVERSATION ON APRIL 15THIn this episode of Coredump, three former Pebble engineers reunite to dive deep into the technical quirks, philosophies, and brilliant hacks behind Pebble OS. From crashing on purpose to building a single codebase that powered every watch, they share war stories, bugs, and what made Pebble’s firmware both rare and remarkable. If you love embedded systems, software-forward thinking, or startup grit— this one’s for you.Key topics:Pebble intentionally crashed devices to collect core dumps and improve reliability.All Pebble devices ran on a single codebase, which simplified development and updates.The open-sourcing of Pebble OS is a rare opportunity to study real, commercial firmware.A platform mindset—supporting all devices and apps consistently—shaped major engineering decisions.Pebble’s app sandbox isolated bad code without crashing the OS, improving developer experience.The team built a custom NOR flash file system to overcome constraints in size and endurance.Core dumps and analytics were essential for tracking bugs, deadlocks, and field issues.Collaborations between hardware and firmware engineers led to better debugging tools and smoother development.Chapters:00:00 Episode Teasers & Intro01:10 Meet the Team: Pebble Engineers Reunite01:13 Meet the Hosts + Why Pebble Still Matters03:47 Why Open-Sourcing Pebble OS Is a Big Deal06:20 The Startup Firmware Mentality08:44 One OS, All Devices: Pebble’s Platform Bet12:30 App Compatibility and the KEMU Emulator14:51 Sandboxing, Syscalls, and Crashing with Grace20:25 Pebble File System: Built from Scratch (and Why)23:32 From Dumb to Smart: The Iterative Codebase Ethos26:09 Core Dumps: Crashing Is a Feature30:45 How Firmware Shaped Hardware Decisions33:56 Rust, Easter Eggs, and Favorite Bugs36:09 Wear-Level Failures, Security Exploits & Font Hacks39:42 Why We Chose WAF (and Regret Nothing?)42:41 What We’d Do Differently Next Time47:00 Final Q&A: Open Hardware, Protocols, and Part Two?Join the Interrupt Slack ⁠⁠⁠⁠Watch this episode on YouTube⁠Suggest a Guest⁠⁠⁠Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website

In today’s Coredump Session, we unpack the full story of Bluetooth—from its PDA-era beginnings to its rising role in cloud-connected devices. With insights from Memfault’s Chris Coleman and François Baldassari, along with Blecon’s Simon Ford, this wide-ranging conversation explores how Bluetooth Low Energy has evolved, where it thrives (and doesn’t), and why it’s often the right tool, even if it’s not a perfect one. Expect history, hot takes, and practical guidance for building better Bluetooth-powered products.Key Takeaways:Bluetooth Low Energy (BLE) and Bluetooth Classic are fundamentally different—and BLE was never just a “lite” version.BLE's strength lies in its low power consumption and quick connection setup, making it ideal for peripheral devices that sleep most of the time.Use cases like audio, asset tracking, and cloud sync continue to shape BLE’s evolution, and new specs like LE Audio and PAwR are expanding its reach.Bluetooth wins not because it’s perfect—but because it’s practical: globally adopted, low-cost, and well-supported.Debugging Bluetooth at scale requires collecting connection parameters, analyzing retries, and understanding phone ecosystem quirks.BLE Mesh adoption has been underwhelming, with real-world complexity often outweighing its theoretical benefits.Expect to see BLE turn up in more places, including MEMS sensors and energy-harvesting devices, not just consumer gadgets.Designers should understand trade-offs in connection intervals, latency, and power draw when choosing Bluetooth for cloud or local connectivity.Chapters:00:00 Episode Teasers & Intro01:10 Meet the Guests: Bluetooth Roots at Pebble, Fitbit, and Blecon06:51 BLE’s Breakthrough: The iPhone 4S Moment10:22 BLE vs Classic: Why It Took Off14:39 Specs That Shifted Everything: Packet Length, Coded PHY & LE Audio21:41 Is BLE Still Interoperable? And Does It Matter?28:22 The BLE Cloud Puzzle: Gateways, Phones & Golden Gate38:40 BLE’s Sweet Spot: Power, Latency & When It Just Works47:12 Operating BLE Devices in the Wild: What to Track & Why57:40 Mesh Ambitions vs Reality⁠⁠Join the Interrupt SlackWatch this episode on YouTube⁠Suggest a Guest⁠Follow Memfault⁠⁠LinkedIn⁠⁠⁠⁠Bluesky⁠⁠⁠⁠Twitter⁠⁠Other ways to listen:⁠⁠Apple PodcastsiHeartRadio⁠⁠⁠⁠Amazon MusicGoodPodsCastbox⁠⁠⁠⁠Visit our website