Update README.md

README.md

@@ -1,10 +1,176 @@
+---
+title: README
+emoji: 🚀
+colorFrom: indigo
+colorTo: yellow
+sdk: static
+pinned: true
+thumbnail: >-
+  https://cdn-uploads.huggingface.co/production/uploads/6634fc18d94421fe1c02f97c/48breLiEtms1xr-xl36dc.png
+short_description: Embedl - efficient AI for the edge
+---
+
+# Embedl
+
+Embedl develops advanced tools and algorithms for **Edge AI**. Our mission is to make AI models run 
+**faster**, **more energy-efficient**, and **reliably across diverse hardware platforms**, while 
+significantly reducing development time.
+
+We help teams deploy high-performance AI on real-world, resource-constrained devices.
+
 ---
-title: README
-emoji: ⚡
-colorFrom: indigo
-colorTo: yellow
-sdk: static
-pinned: false
+
+## Core Products
+
+### **Embedl SDK** *[Enterprise](https://www.embedl.com/sdk)*
+
+A **source-available**, toolkit that supports the **entire Edge AI development lifecycle**.  
+The SDK integrates with **any hardware or toolchain**, including:
+
+- NVIDIA GPUs (Jetson AGX Orin, Nano, Thor, Drive Thor … )
+- Qualcomm & TI accelerators 
+- CPUs, MCUs, FPGAs, and custom accelerators  
+
+The SDK follows Embedl’s **C3PO Framework**:
+
+- **Compatibility** – through model surgery to fix op support issues 
+- **Provisioning** – of hardware access, runtimes, and compilation servers 
+- **Pipeline** – to compile, quantize, and run on-device inference for any model and hardware  
+- **Profiling** – on-device latency, memory, and performance for layerwise statistics  
+- **Optimization** – with public and proprietary algorithms for:
+  - Mixed-precision quantization  
+  - Pruning
+  - Neural Architecture Search (NAS)
+  - Knowledge Distillation  
+
 ---
 
-Edit this `README.md` markdown file to author your organization card.
+### **Embedl Hub** *[Free Beta](https://hub.embedl.com)*  
+
+A cloud-based platform for **quantization, compilation, benchmarking, and deployment** on real edge
+devices (Android & iOS) through device farms.
+
+Key features:
+
+- Run models on real devices via the cloud  
+- Profile latency and performance  
+- Track experiments and compare results across devices  
+- Deploy optimized models directly to edge environments  
+
+---
+
+### **Embedl Visualizer** *(Enterprise)*
+A powerful visualization tool for understanding model performance across the stack.
+
+Supports:
+- PyTorch & ONNX models  
+- Compiled artifacts (TensorRT engines, TIDL graphs, etc.)
+
+Capabilities:
+- Identify latency bottlenecks quickly  
+- Debug QAT issues caused by operator fusion  
+- Compare multiple models and configurations  
+- Track how a model evolves across compilation stages — from Python code to final deployable binaries  
+
+---
+
+### **Embedl Models** ([Community](https://huggingface.co/embedl))
+
+Pre-optimized models that can be used **off-the-shelf** or customized for specific hardware target
+supported by the [embedl-models](https://github.com/embedl/embedl-models) package.
+
+**First release highlights:**
+
+- The **fastest Small Language Models (SLMs)** using **[FlashHead](https://www.embedl.com/knowledge/ultra-efficient-llms-embedls-breakthrough-for-on-device-ai)**,
+  a novel architectural improvement to the language-model head
+- Works with popular models like **Llama, Gemma, and Qwen**
+- Provides speedups on top of:
+  - Quantization  
+  - Flash Attention  
+  - Other standard optimizations
+
+Device: Nvidia Jetson Thor
+| Model                                            | Generation speed (tokens/s) |
+| ------------------------------------------------ | ----------------------------|
+| embedl/Llama-3.2-3B-Instruct-FlashHead-W4A16     | 100                         |
+| Llama-3.2-3B-Instruct-W4A16*                     | 80                          |
+| RedHatAI/Llama-3.2-3B-Instruct-FP8               | 64                          |
+| meta-llama/Llama-3.2-3B-Instruct                 | 37                          |
+
+*Embedl quantized model for benchmarking similar to the FlashHead-W4A16 but without
+the faster FlashHead and custom generation loop.
+
+---
+
+## Why It Matters
+
+- Enables AI deployment on **resource-constrained hardware**:
+  - Embedded systems  
+  - Mobile devices  
+  - IoT  
+  - Robotics  
+- Reduces **latency, memory usage, and energy consumption**, enabling real-time inference without 
+  cloud dependence  
+- Saves development time through a **hardware-agnostic workflow** reusable across models and platforms  
+- Bridges the gap between **academic ML research** and **industrial embedded AI applications**
+
+---
+
+## Company Information
+
+- **Founded:** 2018 (spin-out from Chalmers University of Technology)  
+- **Commercial Operations:** Since 2022  
+- **Headquarters:** Gothenburg, Sweden  
+- **US Office:** Palo Alto, California  
+- **Recognition:**  
+  - Named to **CB Insights “AI 100” (2025)** list of the most promising private AI companies  
+
+---
+
+## Typical Use Cases
+
+Embedl is used where **real-time performance and efficiency are critical**:
+
+- **Automotive & Autonomous Systems**
+  - Autonomous driving
+  - ADAS
+  - Driver monitoring
+  - Predictive maintenance  
+  - *Example: Kodiak Robotics*
+
+- **Defense & Aerospace**
+  - Secure, energy-constrained AI inference  
+  - *Example: Airbus*
+
+- **Mobile & Edge AI**
+  - Running deep-learning models directly on phones and embedded devices  
+  - No cloud dependency via Embedl Hub  
+
+---
+
+## How to Get Started
+
+### **Quick Start with Embedl Hub**
+- Upload your model (PyTorch or ONNX)  
+- Quantize, compile, and benchmark on supported devices  
+- No physical hardware required  
+
+### **Full Control with Embedl SDK**
+- Integrate Embedl directly into your training and deployment pipeline  
+- Works with TensorRT, QNN, TIDL, and more  
+- Access advanced hardware-aware optimization and performance insights  
+- Deploy to your own infrastructure  
+
+### **Custom & Enterprise Needs**
+For tailored optimizations, specialized hardware support, and engineering collaboration, contact 
+Embedl for full SDK access and support.
+
+---
+
+## Contact
+
+**Headquarters (Sweden)**  
+Gamla Almedalsvägen 39  
+412 63 Gothenburg, Sweden  
+
+**Email:** info@embedl.com