Nokia ECE's (Enterprise Campus Edge) Intel EMF 2025 Adoption – A Retrospection: From Exploration to Execution

[shanki1107]

Our Community-Driven Path to Production-Ready Edge AI

We at Nokia ECE started working with Intel EMF in early 2025 through regular weekly and monthly syncs. The initial goal was simple: understand the platform properly and get hands-on with it rather than staying at a conceptual level.

Very quickly, this turned into real implementation work.

We brought up Edge Manager, validated PXE discovery of our ruggedized AI Far Edge box for Industry 4.0 customers OnLogic Karbon 804 industrial grade GPU enabled boxes powered with NVIDIA Ada RTX2000 GPU and testing the provisioning flows, deployed applications, and exercised platform capabilities like observability, RBAC, monitoring, alerts, SMTP integration, and event workflows. Most of this happened through practical experimentation—deploying, breaking things, debugging, and iterating.

One of the more defining moments for us came during application troubleshooting. We expected to rely on Grafana but found that what we needed wasn’t available in the way we assumed. That pushed us to engage more deeply with the community. We raised issues, worked closely with the Intel team, and started looking at the problem with more rigor.

With their inputs, we traced the issue to a design behavior around the “Keep-Alive” mechanism that was impacting pod stability. This led to focused debugging on our side involving two senior developers and our lead architect, with Intel staying engaged across multiple discussions. We got the fix working within a week, soaked it for two more weeks to ensure stability, and then submitted the merge request upstream.

That flow may sound routine to anyone familiar with open source, but for us it was happening while we were under pressure to finalize our Edge strategy and hit delivery milestones. The fact that we could still contribute upstream while meeting internal timelines speaks volumes about how effective the collaboration was.

What stood out throughout the journey was how much engineering effort EMF saved us. Instead of spending cycles rebuilding core manageability and infrastructure capabilities, we could focus on building meaningful differentiation—like extending UI features and refining workflows for our use cases. At one point, when Intel mentioned they were surprised by how quickly we moved to readiness (around seven months), we were honest in saying: strong team, but the platform and ecosystem played a big role in accelerating us.

There were also real-world constraints to deal with. Our cloud deployment model differed significantly from the default assumptions. The Intel team didn’t treat this as a limitation—they brought in the right experts, went through our provider specifics, project structure, and workflows, and worked with us to identify practical deployment recipes.

We also had a late requirement to validate SMTP integration due to customer expectations. Under time pressure, we worked closely with the community and Intel engineers and managed to get deployment and validation done within a day. Most of it worked as expected; the rest was mainly about stabilizing and wrapping it for our internal context.

After delivering and demoing our product to customers, we took some time to reflect. Looking back at our early demos, learning artifacts, and the photos from the Intel EMF Bangalore event—where Jaisuthan represented our group—we felt this was a journey worth documenting.
That’s why we’ve put together some pictures and a small video showing our PXE onboarding of edge devices, showing the progression from early learning to building real features on top of EMF, and we’re sharing it here along with Intel EMF Bangalore event photos, some pictures with us working with it, and some testing it out in our labs, and of course, having fun together, with food!

We’re thankful for the depth of technical engagement, transparency, and consistency from the Intel EMF team and the wider community. We plan to stay actively involved, continue contributing, and grow alongside this ecosystem.

Nokia-With-IntelEMF-Onboard.mp4

[amathan]

We’d like to sincerely thank the EMF team and community for the continuous support throughout our journey. Your responsiveness in answering our questions, helping us troubleshoot issues, and guiding us during debugging made a real difference for us.

What we appreciated most was not just the technical expertise, but also the willingness to listen, understand our use cases, and work with us to find practical solutions. That level of collaboration greatly accelerated our progress and helped us move forward with confidence.

Grateful to be part of such a supportive open-source community.

[biapalmeiro]

Grateful to have you here.

[scottmbaker]

Aravinth and Shankar, I'd like to thank you and your team for the collaboration. It's been a pleasure participating in the weekly sync calls and seeing the product develop. The use cases and the issues you have brought have helped us to make the Edge Manageability Framework better and it's great to see the visuals of your team enthusiastically working on the product. Looking forward to more collaboration in the future and seeing the capabilities grow and develop.

[biapalmeiro]

I love experiencing this journey through your perspective, @shanki1107. Seeing results like these brings joy to the entire journey we share. The energy and passion conveyed through the photos embody the very spirit of open-source—very human and very inspiring. Thank you!

[shanki1107]

Thank you biapalmeiro for letting us be part of this journey.

[Amithpn]

We took a dare step of adopting EMF and it was a landmark decision. Saved months of our effort and with such efficiency we really had to spend most of the time testing feature after feature. Mind boggling. Thank you entire EMF team. Great support. Wonderful journey, great collaboration. Many more to come.

[biapalmeiro]

That is exactly what we want to deliver: an easy development journey. You focus on your IP—we handle the edge complexity. ;)

[BasuGoudar-GitHub]

Thank you @Amithpn, @shanki1107, @amathan for the feedback and happy to see that Intel EMF Solution helped Nokia to scale MXGrid solution. This is a true testament to engineering capability and deliverables by the EMF Team. We appreciate the Nokia team's technical leadership and collaboration working with us without boundaries. Let's continue this collaboration.

[jaisuthan]

I recently attended the “Industry Solution Builders Connect: Unlocking Business Value with Edge AI” event organized by Intel on October 7, 2025, at Radisson Blu, Bengaluru. It was a power-packed and insightful event featuring several key speakers from Intel and across the industry, focused on the evolving landscape of Edge AI.

The sessions provided a comprehensive view of Intel’s latest hardware platforms and software stack for Edge AI, highlighting their capabilities in accelerating real-world AI deployments. The event covered several important themes, including:

Dynamic Edge AI Roadmap: Shaping Tomorrow

Developing Solutions on Intel Edge AI Systems

Accelerating Edge AI with Intel Platforms

The Evolution of Edge AI Over the Last Decade and the Current Inflection Point

There was also a strong focus on leveraging Intel hardware and software for Generative AI at the edge, which provided valuable insights into emerging use cases and deployment considerations.

The event featured engaging discussions and Q&A sessions with industry leaders, including a panel discussion with the Managing Director of Bits and Bytes (BBIPL). One of the notable keynote sessions was delivered by Santhosh Vishwanathan, Vice President and Managing Director – India Region, Intel, who spoke on the “Age of Intelligence” and shared valuable perspectives on industry trends and the future direction of AI adoption.

In addition, several Intel partner stalls showcased their products and solutions. This provided a great opportunity for hands-on interactions and technical discussions. Many partners demonstrated computer vision-based applications similar to our Visual Positioning and Object Detection solution, which helped spark new ideas and perspectives for our own product roadmap. The networking opportunities were particularly valuable—I was able to establish new contacts with members of the Intel Bengaluru team as well as with companies working on similar problem domains.

Some of the notable partners included SchoolNET, Acer, SwitchON, VVDN, and Roombr.

Overall, the event was highly informative and well-organized, offering a strong blend of strategic vision, technical depth, and networking opportunities. It provided meaningful insights into the current state and future direction of Edge AI and reinforced the growing importance of Intel’s ecosystem in enabling scalable, real-world AI solutions at the edge.

[shanki1107]

Thanks @jaisuthan for sharing your experience. It was he who represented our entire group at the event and I have added a picture of him at the stalls there checking out the product capabilities as well. Jai and his team are actively working on the next pillar of our engagement with Intel, the video analytics platform. Looking forward!

[amul1umt]

Such a great testament to open source efforts! Thank you, Nokia team!! Amazing collaboration!