From Sensor to Inference
Macnica AmericasCapture. Process. Communicate.
Live panel webinar
Holoscan Panel Series

From Sensor to Inference

Building a complete edge AI pipeline, from the physical world to the decision.

Tuesday, November 3, 2026 11:00 AM Central 60 minutes Live, with Q&A
Register your interest
The premise

Edge AI is only as good as the data pipeline feeding it

Imaging often dominates the bandwidth discussion around Holoscan. Real physical AI systems increasingly combine cameras with radar, time of flight sensors, and multi microphone arrays. Each sensor type brings its own interface, bandwidth, synchronization, and preprocessing needs before a single byte reaches the GPU.

This panel follows one event through one system. An autonomous mobile robot is moving down a busy industrial aisle. A person steps out from behind a rack. What has to happen, and how fast, before the robot stops?

What you will learn

Three takeaways for your next design

01 Sensing

How different sensor types affect bandwidth, latency, and synchronization in an edge AI pipeline.

02 Data path

Where FPGA based aggregation and preprocessing can improve the data path to GPU based inference.

03 Deployment

How to move from a sensor and AI proof of concept to a deployable Holoscan platform.

Agenda

One system on screen, one block lighting up at a time

Capture Process Whole system
  1. 5 min

    The scenario and the system

    Meet the robot, the event, and the system diagram: sensors, data plane, compute, decision. The question for the hour: what has to happen, and how fast, between that person appearing and the robot stopping?

    Macnica Americas
  2. 10 min

    Detect: what the system sensesCapture

    Which sensors catch this event, and why you need more than one. Radar for motion in poor visibility, time of flight for near field depth, the microphone array for sound cues, imaging for classification. The data each one produces, its rate, and its interface.

    Infineon
  3. 10 min

    Align: turning four streams into one pictureProcess

    The same four streams, arriving over different interfaces with different timing. How an FPGA converts interfaces, timestamps and synchronizes, preprocesses, aggregates, and moves the result efficiently toward the GPU. Where 10G, 25G, and 100G tiers fit.

    Altera
  4. 10 min

    Decide: running perception on aligned dataProcess

    The same data arriving at a Jetson based Holoscan platform. How it is ingested and processed, and what changes when a bench prototype becomes a robot: connectivity, power, thermal, mechanical, and I/O.

    Connect Tech
  5. 15 min

    Stress test the system

    One change at a time, same system. Each panelist says how their layer responds.

    • Camera resolution doubles
    • A second radar is added
    • One sensor stream drops out or drifts out of sync
    • The platform must fit a smaller, hotter enclosure
    Macnica Americas moderates
  6. 10 min

    Audience Q&A and close

    Your questions to the panel, plus resources and next steps on the Macnica Holoscan landing page.

    All panelists
Your panel

Four teams, one architecture

Photo of Kim Lee
Infineon

Kim Lee

Sr. Director, System Applications Engineering Detect

Kim is responsible for the global deployment of Edge AI and IoT applications across Infineon's mass market channels, driving the enablement that helps customers accelerate product development with Infineon solutions. Throughout his 20+ year Infineon career, he has held leadership roles spanning regional marketing, systems engineering, and applications engineering across technologies including RF/mmWave, communications, sensors, IoT, and power systems. His career also includes engineering and marketing leadership positions at Lantiq/Intel, Microchip Technology, and Sipex Corporation.

Photo of Seong H. Kim
Altera

Seong H. Kim, Ph.D.

Associate VP, Solutions Architecture Align

Seong H. Kim, Ph.D., is an Associate VP at Altera with more than 30 years of experience in communications, cloud, data center and physical AI solutions. He specializes in defining and delivering accelerated computing architectures across FPGA, GPU, CPU, NPU, Arm, and x86 platforms, with a focus on AI, compute, networking, SmartNIC, storage and robotics applications.

Throughout his career, Seong has led global solutions architecture teams, developed customer focused proof-of-concept solutions, and managed an ecosystem of more than 200 technology partners. He has worked with leading cloud service providers, technology companies, telecommunications equipment manufacturers, and enterprise customers to identify application bottlenecks and build differentiated solutions that improve performance and total cost of ownership.

Seong holds a Ph.D. in Electrical and Computer Engineering and an MBA in Marketing, combining deep technical expertise with a strong business perspective. He has authored more than 30 technical papers, white papers, and articles, delivered numerous industry seminars and keynotes, and holds 12 patents.

Photo of David Pearson
Connect Tech

David Pearson

Support Applications Engineer Lead Decide

David is Support Applications Engineer Lead at Connect Tech Inc., an NVIDIA Elite Partner, and host of Connect Tech Talks, the company's technical video series on Edge AI and robotics platforms. A computer science graduate of the University of Guelph, he leads applications support for Connect Tech's Edge AI portfolio and is the company's recurring on-camera technical voice, explaining NVIDIA Jetson Thor, IGX, and DGX Spark platforms, digital twin and Real2Sim workflows, and rugged deployment for robotics, defense, and heavy industry.

Photo of Chris Cook
Macnica Americas

Chris Cook

Field Applications Engineering Manager Moderator

Chris is a technology and business development professional with more than 20 years of experience in FPGA technology, system architecture, and embedded computing. With a background working with Altera and Xilinx FPGA technologies, Chris brings a system-level perspective to solving complex design challenges and helping customers turn innovative technologies into practical solutions. At Macnica Americas, Chris focuses on FPGA-based architectures and emerging AI applications, particularly edge AI systems that integrate sensors, programmable logic, and GPU computing. He is passionate about how FPGA technology can enable efficient data movement, real-time processing, and scalable AI architectures from sensor to inference.