embedUR systems spins out ModelNova to accelerate Edge AI development lifecycles

Chennai, July 17: ModelNova™, an Edge AI firm spun out of embedUR systems, unveiled its full‑scale launch as an independent entity focused on bringing practical AI to low‑power edge devices. 

As the industry accelerates toward “physical AI” — bringing intelligence out of the cloud and into the real world so that machines can sense, decide, and act on their own — the biggest challenge is no longer training a model but ensuring it runs reliably on constrained, AI-optimized silicon. ModelNova was created to bridge that gap by pairing ready‑to‑use building blocks with the tooling and expertise required to move models all the way to high‑volume production. 

“Every technology wave succeeds or fails depending on whether developers can actually build with it, and Edge AI is now at that tipping point. The distance between a functional model and a shipped product is what’s restraining the industry — ModelNova targets exactly that issue, and its launch marks a significant advance for the whole ecosystem,” said Michael Hurlston, President and CEO, Lumentum. 

One company, the complete Edge AI lifecycle 

ModelNova’s offering is built around three complementary pillars: 

Model Zoo — A free, open‑source library of pretrained models, datasets, and use‑case blueprints tuned for tiny, low‑power devices. The Zoo lets developers kick‑start a proof‑of‑concept on real hardware—from a Raspberry Pi to AI-native silicon from ModelNova’s semiconductor partners, who include ARM, STMicroelectronics, Infineon, Alif Semiconductor, NXP, Silicon Labs, Renesas, Synaptics, and Ceva. ModelNova is also a Leadership Partner of the EDGE AI FOUNDATION, the global community advancing Edge AI research, education, and adoption. 

Fusion Studio — A commercial development platform that guides the entire journey from model to product across MCU and NPU silicon. While many tools stop after model compilation, Fusion Studio provides end‑to‑end system integration—managing the pipeline, runtime, and the silicon‑specific tuning needed to turn a promising prototype into a market‑ready product. 

Licensable production-ready models — Five hardened, production-ready models. NovaEyeD (facial recognition) is designed for security‑critical uses like access control, keyless entry, and personalization, while a set of wireless‑sensing models supports everything from motion and micromotion detection to automotive child‑presence detection over Wi‑Fi, UWB, mmWave, and beyond. 

“The Alif Ensemble Series was created to deliver real AI inference on low‑power silicon, and ModelNova has demonstrated what’s achievable on it—from free Zoo models to production‑ready deployments tuned for our MCUs. Their spin‑out as an independent firm gives developers a clearer route from idea to product, which is precisely what this partnership aims for,” said Reza Kazerounian, Co‑Founder & President at Alif Semiconductor. 

Founded on two decades of embedded expertise 

ModelNova’s launch as an independent company follows 18 months of momentum. Its model zoo already drives Edge AI ecosystems for two leading silicon vendors, with production‑ready models validated in real‑world consumer, medical, and AIoT applications. 

“ModelNova started as our answer to a simple question: why is it still difficult to turn a working AI model into a real‑world edge product? By making ModelNova an independent company, we can focus entirely on that challenge—offering developers a free starting point, a development platform, and licensable production‑ready models,” said Rajesh Subramaniam, CEO & Founder of embedUR systems. 

“When we launched the EDGE AI FOUNDATION, we believed that bringing silicon vendors, researchers, and developers together around shared knowledge would spur real companies to solve actual deployment challenges. ModelNova’s spin‑out as an independent firm—covering open‑source models, professional tools, and production licensing—is precisely the maturation we envisioned. It’s a proud milestone for the entire Edge AI community,” said Pete Bernard, CEO of the EDGE AI FOUNDATION.