ipXchange

Find the OFFICIAL LIVESTREAM for Edge Impulse's Imagine Innovators here:

Its's Live Stream Day! We are here in Amsterdam at Imagine Innovators by Edge Impulse. We are live streaming technical sessions all day, presented by Brian McFadden & Adam Yap. See the live stream session agenda here https://luma.com/majggd1w. Live stream: https://lnkd.in/e7_Fdjkh #EdgeAI #EmbeddedAI #EmbeddedMachineLearning #TinyML #MachineLearning #ArtificialIntelligence #EmbeddedSystems #DeveloperCommunity #Robotics #ROS2 #Linux #EdgeComputing #Hardware #AIEngineering #TechEvent

Last chance to register! Join Foundries.io and Edge Impulse as they walk through a practical workflow for building production IoT systems. This webinar is designed to bridge the gap between knowing how to train and optimise AI models, and having a clear, secure, and repeatable path to deploying them on embedded Linux devices in production. What you’ll learn - Show how engineers can navigate the complexities of Linux infrastructure for edge AI. - How to use containers and embedded Linux to create reliable, production-ready IoT devices. - Understanding the infrastructure: CI/CD pipelines, containerisation, Bill of Materials (BOM), secure OTA. - How customised and secure AI models can be trained and moved into a deployment-ready Linux container. - A live demonstration of the FoundriesFactory architecture. Register here: https://lnkd.in/emip5huS #EdgeAI #EmbeddedLinux #ProductionIoT #Containerisation #OTAUpdates #DeviceManagement #SecureIoT #EmbeddedSystems #AIDeployment #FoundriesIO #EdgeImpulse #IoT Qualcomm oemsecrets.com

The Engineer’s Guide to Industrial Edge AI: Rightsizing Hardware for Multi-Model Vision and SLM Applications This webinar brings together OnLogic, Edge Impulse and Qualcomm to discuss how engineering teams can build more practical, production-ready edge AI systems for industrial and manufacturing applications. What you’ll learn: - What engineers need to consider when moving from prototype to deployment - Why you might want to use less powerful AI models or compute hardware - How to deploy Multi-model vision onto industrial hardware - How to choose what model is active in multi-model systems - How to balance compute, power, latency, accuracy, cost and reliability - Where smaller models, SLMs and agentic AI workflows fit within an industrial AI stack - How edge AI can support inspection, monitoring, quality control and automation - How hardware and software choices affect long-term scalability and production readiness 🗓️ 10th of June, 4pm BST Register here: https://lnkd.in/erdxEt7A #IndustrialAI #EdgeAI #VisionAI #ManufacturingAI #MachineVision #IndustrialAutomation #ComputerVision #AIDeployment #ProductionAI #OnLogic #EdgeImpulse #Qualcomm

What makes AI inference faster and more efficient at scale? Our conversation with Sayan M. from SambaNova, Sandro looked at their Reconfigurable Dataflow Unit - a proprietary silicon architecture designed for the AI agentic era. Unlike traditional GPU execution, the RDU uses a dataflow-based execution model and a three-tier memory structure to reduce kernel-by-kernel overhead, improve inference performance, and lower power consumption. SambaNova also offers cloud access and API keys for developers who want to evaluate the platform directly. #AI #LLM #Inference #Semiconductors #Dataflow #SambaNova #AIInfrastructure

While we were over at Austin, we decided to pop into the local Tenstorrent office for them to show us around. Miles Dooley was kind enough to sit down and discuss what they actually get up to in there beautiful offices, what technologies they have created and how this might impact engineers in the larger world. After such a insightful conversation, it was only right that we had a peak around their labs to see this technology in action, Including a demo they prepared for us showcasing how they test hardware before it even physically exists by leveraging FPGA's flexible circuitry. Luckily we had Brandon Zupan to share his technical insight into many of the devices and techniques that are developed and used in their labs, including their testing of the upcoming Tenstorrent Ascalon S. #EdgeAI #EmbeddedAI #ArtificalIntelligence #Tenstorrent #TTAscalon #QuietBox #FPGA #ChipDesign #ProcessorDesign #DesignEngineering #HardwareEngineering #ipXchange

You might not see Ambiq front and centre, but its technology is invisibly enabling devices that we wear to have the breadth of functionality that we have come to expect from wearables today. Their Subthreshold Power Optimised Technology (SPOT) technology has been the cornerstone of their ultra low power silicon. This is no doubt a great achievement in a world where low power operation is pivotal for always-on, battery-limited applications. While we were in Austin, we spoke with Dr. Adam Page, Head of Artificial Intelligence (AI) at Ambiq, about the company’s ultra-low-power approach to edge AI for wearables and battery-powered devices. Dr. Adam Page shared more than SPOT, they had kindly prepared 3 demos for us, showing how its Apollo 510 platform can run multiple AI models on-device. Interestingly enough the demos showcased the company's expertise in reducing power requirements. One of the most interesting parts was the compression demo. For wearable devices collecting ECG, PPG or inertial measurement unit (IMU) data, transmitting everything to the cloud can consume a lot of power. Ambiq’s compression approach is designed to reduce the amount of data that needs to be stored or transmitted, while keeping the important signal information intact. For engineers building smart rings, health monitors, hearables, smart glasses or other low-power edge AI devices, It is not just about having low-power silicon. It is about combining the hardware, runtime, software tools and model deployment workflow so that AI can actually run efficiently at the edge. Watch the video to find out more about Ambiq's mission. #EdgeAI #EmbeddedAI #Wearables #LowPowerDesign #Microcontrollers #ArtificialIntelligence #IoT #EmbeddedSystems #SmartWearables #HealthTech

Technical Developer Livestream! Edge Impulse is bringing Imagine Innovators to Amsterdam on May 20th. If you are not able to attend in person, don’t worry, the full day will be live streamed. The event brings together senior engineers, developer advocates, AI leads, solution architects, researchers, product leaders and technology experts working at the forefront of embedded machine learning. Sessions will run from 10 am to 6 pm, so you can jump in and watch the sessions that interest you most. You’ll see: - A ROS 2 robot inspection deep dive - A production Linux edge AI discussion with Foundries.io - A live model poisoning and repair demo - Live AI agent project building - A Reachy Mini live Q&A - Hardware breakdowns - Doom running on hardware - An Edge Impulse R&D roundtable covering platform capabilities, roadmap and behind-the-scenes development Register & see session times here: https://luma.com/majggd1w #EdgeAI #EmbeddedAI #EmbeddedMachineLearning #TinyML #MachineLearning #ArtificialIntelligence #EmbeddedSystems #DeveloperCommunity #Robotics #ROS2 #Linux #EdgeComputing #Hardware #AIEngineering #TechEvent

Deploying AI models to microcontrollers often involves trial and error, especially when it comes to memory constraints and inference performance. Ernest Scholtz form Deepgate explains to Sandro how their platform designed to simplify that process. We also explore real use cases such as intelligent sensing, where edge AI filters data locally and only sends anomalies to the cloud, reducing bandwidth and cost. To evaluate this technology yourself visit: https://deepgate.ai/ #EdgeAI #TinyML #IoT #MachineLearning #AIDeployment #EmbeddedSystems #MLOps #IntelligentSensing