Imagine
Innovators

The developer conference for edge AI

May 6, 2025 8:00 am PT- 11:00 am PT

Join us for a fully virtual developer conference about edge AI. Hear the latest in edge AI from leading technical experts, take hands-on tutorials, and engage with the developer community.

Why attend?

Connect

with developers

Hear the latest advances in edge AI, engage in a technical session, join a workshop, and connect with your peers.

Engage

with AI experts

Connect with experienced AI experts and learn best practices for building, optimizing, and deploying edge AI solutions at scale.

Keynotes/Panels

Keynotes/Panels
8:00 am
-
8:30 am
Keynotes/Panels
8:00 am
-
8:30 am
PST

Welcome Keynote

Speaker:

Zach Shelby

Keynotes/Panels
8:30 am
-
9:00 am
Keynotes/Panels
8:30 am
-
9:00 am
PST

Edge AI use cases

Host: Mike Senese

Participants:

  • Marta Barbero, Lead Product Manager at Arduino
  • Pete Bernard, CEO of the EDGE AI FOUNDATION
  • Olivier Bloch, IoT advisor and host of the IoT Show
  • Kedar Gharat, Product Director of AI Solutions at Qualcomm
Keynotes/Panels
9:00 am
-
9:30 am
Keynotes/Panels
9:00 am
-
9:30 am
PST

Frontier between Generative AI and Edge AI

Host: Sergi Mansilla

Participants:

  • Daniel Situnayake, Director of ML at Edge Impulse
  • Vijay Janapa Reddi, Ph. D., Associate Professor at Harvard University
  • Steve Nouri, CEO and Founder at GenAI Works
  • Dr. Vinesh Sukumar, VP - Head of Gen AI/ML Product Management at Qualcomm
Keynotes/Panels
9:30 am
-
10:00 am
Keynotes/Panels
9:30 am
-
10:00 am
PST

Edge AI at Qualcomm

Host: Louis Moreau

Participant:

  • Manny Singh (VP of Product Management - Industrial and IoT business at Qualcomm)
Keynotes/Panels
10:00 am
-
10:30 am

Amir Sherman

The evolution of Edge AI semiconductors: Tracing their past, shaping their present, and envisioning their future

In this presentation we will explores the evolution of EdgeAI semiconductors, examining their innovations from the past, advancements in the present, and their transformative potential for the future. It highlights the pivotal role of microcontrollers (MCUs), microprocessors (MPUs), and system-on-chips (SoCs) in enabling intelligent, decentralized computing, shaping industries, and overcoming emerging challenges.

Keynotes/Panels
10:00 am
-
10:30 am
PST

The evolution of Edge AI semiconductors: Tracing their past, shaping their present, and envisioning their future

Speaker:

Amir Sherman

Keynotes/Panels
10:30 am
-
11:00 am
Keynotes/Panels
10:30 am
-
11:00 am
PST

Closing Keynote

Speaker:

Jan Jongboom

Beginner Track
Beginner Track
8:30 am
-
9:00 am

Jim Bruges

Get Started With Edge Impulse

Learn about the fundamentals of Edge AI and how to start building your first model from data collection to deployment. In this introductory session you'll get an overview of how Edge Impulse helps any engineer create actionable insights on their edge devices through simple model training tools and seamless deployment.

Beginner Track
8:30 am
-
9:00 am
PST

Get started with Edge Impulse

Speaker:

Jim Bruges

Beginner Track
9:00 am
-
9:30 am

Louis Moreau

Datasets best practice: From prototype to production

Data are everywhere, but how do you build robust datasets to train solid Edge AI models? In this session, you'll learn how to create reliable datasets for Edge AI applications. Louis Moreau will guide you through starting with just a few hundred samples to validate initial ideas and scaling up to construct robust data pipelines for production use cases. You'll understand why using robust and well-labeled datasets is crucial for model performance, and some tips and tricks to ease your data collection and data labeling processes.

Beginner Track
9:00 am
-
9:30 am
PST

Datasets best practice: From prototype to production

Speaker:

Louis Moreau

Beginner Track
9:30 am
-
10:00 am

David Tischler

Introduction to Edge MLOps

Engineers and developers tasked with deploying AI models out in to the field face a variety of unique challenges. One of the most important considerations is the creation of a reliable and repeatable feedback loop, allowing new data to be collected in order to further improve model accuracy. This talk will cover some of the popular solutions and strategies that can help establish an MLOps process and workflow, including fleet management, connectivity, and OTA methods. These functionalities then enable data to be collected from deployed devices, ultimately resulting in more robust models delivered back to edge hardware.

Beginner Track
9:30 am
-
10:00 am
PST

Introduction to Edge MLOps

Speaker:

David Tischler

Beginner Track
10:00 am
-
10:30 am

Minhea Stoica

Smart Sensors for Smarter Robots: Improving Robotic World Understanding Through Edge AI

This presentation explores how robots can gain a better world understanding and interact with their surroundings using Sensor Fusion technology. The discussion covers two practical examples: an electronic nose that distinguishes between different smells, and a touch-sensing system that helps robotic hands confirm their grip. These projects demonstrate how combining multiple sensors with Edge ML and micro-ROS creates smarter robotic systems.These examples illustrate how enhanced sensing abilities make robots more capable and reliable partners across various applications.

Beginner Track
10:00 am
-
10:30 am
PST

Smart Sensors for Smarter Robots: Improving Robotic World Understanding Through Edge AI

Speaker:

Mihnea Stoica

ML Practitioners Track
ML Practitioners Track 
8:30 am
-
9:00 am

Brian McFadden

Training novel architectures: the power of custom learning blocks

New machine learning architectures are emerging every day, and you may have even developed your own unique architecture for a specific application. But how can you leverage the effectiveness of the Edge Impulse platform while exploring these latest innovations? In this session, Brian will demonstrate how to use custom learning blocks to seamlessly integrate novel architectures into your impulse, empowering you to test and refine your custom models with ease.

ML Practitioners Track 
8:30 am
-
9:00 am
PST

Fine-tune custom model architecture with your own data

Speaker:

Brian McFadden

ML Practitioners Track 
9:00 am
-
9:30 am

Moe Sani

FOMO-AD: Visual anomaly detection model for industrial use cases

This session will explore FOMO-AD, a cutting-edge approach to visual anomaly detection designed specifically for industrial use cases. We'll delve into how Edge Impulse's visual anomaly detection model operates—covering everything from configuring and training the model to tailoring it for end-of-line quality control in production lines. Attendees will gain insights into the challenges of detecting subtle, unpredictable defects that often escape human oversight and learn how to automate these critical inspections to achieve more reliable and scalable defect detection. Join us for a 30-minute deep dive into building, deploying, and optimizing anomaly detection systems that enhance product quality and streamline industrial processes.

ML Practitioners Track 
9:00 am
-
9:30 am
PST

FOMO-AD: Visual anomaly detection model for industrial use cases

Speaker:

Moe Sani

ML Practitioners Track 
9:30 am
-
10:00 am

Alex Elium

Why bother with DSP for feature extraction?

If the promise of machine learning is that a model can learn any output given training data, then why do engineers bother with feature extraction, especially complex DSP algorithms.  In this talk, we’ll explore a model trained with and without feature extraction against real world data, which will allow us to compare and contrast outcomes.  We’ll show differences in footprint, which is often sited as an advantage for DSP based feature extraction.  Also, we can quantify an often overlooked advantage…a model will only work without feature extraction given enough training data.  We will quantify how much less training data the same model needs when using DSP.

ML Practitioners Track 
9:30 am
-
10:00 am
PST

Why bother with DSP for feature extraction?

Speaker:

Alex Elium

ML Practitioners Track 
10:00 am
-
10:30 am

Ashvin Roharia

Generate & Label Synthetic Data Using AI

"Your AI is Only as Good as Your Data" but collecting and labeling data is often difficult, time-consuming, and expensive. Join us to learn how to generate synthetic data (images or time-series) within Edge Impulse and how to label that entire dataset with our AI labeling tools.

ML Practitioners Track 
10:00 am
-
10:30 am
PST

Synthetic data and AI labeling

Speaker:

Ashvin Roharia

Running on the Edge Track
Running on the Edge Track
8:30 am
-
9:00 am

Andy Doan

The Challenges of Updating Models at the Edge

There are a few different ways to update models for fleets of devices. This session will cover some of them, discuss their pros and cons, and explain an approach we’ve used at Foundries.io.

Andy is a principal engineer at Qualcomm managing the Foundries.io Cloud and OTA teams. Prior to that he was a software engineer at Canonical and Linaro working on various cloud infrastructure projects related to the Arm ecosystem.

Running on the Edge Track
8:30 am
-
9:00 am
PST

The Challenges of Updating Models at the Edge

Speaker:

Andy Doan

Running on the Edge Track
9:00 am
-
9:30 am

Vojislav Milivojevic

Edge AI: milliwatts to value

Join us for a webinar designed for embedded engineers exploring Edge ML. Whether you’re starting out or experienced, this session delves into essential aspects of ML on microcontrollers and firmware development. We’ll discuss practical ML tasks like building datasets and running inference on your Nordic device. Gain insights into using advanced models in Edge ML development, and understand whether specialized hardware is necessary for deploying such models effectively.

Running on the Edge Track
9:00 am
-
9:30 am
PST

Edge AI: milliwatts to value

Speaker:

Vojislav Milivojevic

Running on the Edge Track
9:30 am
-
10:00 am

Jim Bruges

From Tiny ML to Gen AI at the Edge

This session will explore how increasing compute power at the Edge is enabling faster, more complex and more reliable Edge AI solutions. Attendees will explore practical methodologies for combining low power and high power Edge AI models in cascades to create the most efficient and reliable solution to real industrial problems; including a look to the future of Edge AI- how Generative AI can be used to solve problems in real production scenarios.

Running on the Edge Track
9:30 am
-
10:00 am
PST

From Tiny ML to Gen AI at the Edge

Speaker:

Jim Bruges

Running on the Edge Track
10:00 am
-
10:30 am

Eoin Jordan

Running on Android / WearOS

Android devices come in all shapes and sizes, from wearables and embedded devices to automotive consoles and VR headsets, offering a wide range of sensor input. In this session we will look at integrating a new sensor on Android / WearOS project using Edge Impulse. Attendees will learn how to add an additional sensor to a project collect data and deploy back to the given hardware. Demonstrating the end-to-end workflow for working with a new type of Android device and sensor data with Edge Impulse

Running on the Edge Track
10:00 am
-
10:30 am
PST

Running on Android WearOS

Speaker:

Eoin Jordan