Company Performance Metrics
- Chien-Wen Huang: Machine Learning Engineer
- Jay Chen: Deep Learning Solution Archiector
Avalanche Computing is a deep tech provider, which provides hyper-scale computing technology for the intelligence industries. Since 2018, their founder work for NVIDIA USA and did performance analysis for AI applications in data center applications or large-scale applications (FB, Google, AWS, MS, and et., al.). Now, they focus on providing the
performance-optimized AI workflow for the smart medical industry (Medical AI) and smart manufacturing at the beginning. After 2021 CES, they announced the FAST-AI, a clicks-through deep learning accelerator tool that support more than 3 verticals.
The mission of Avalanche Computing is to help AI experts and domain experts saving time on the following tasks: 1. No AI environment configuration 2. AI model training without changing the original network design 3. AI model inference as fast as possible, no deployment pipeline configuration 4. Help the traditional company to use AI in clicks which saving 50% hardware cost and more than 7 times faster than before 5. Providing the best performance-optimized AI workflow (end-to-end solution) to the privacy-sensitive clients (who cannot use the public cloud)
Such that the AI experts can focus on their innovation. Now, Avalanche Computing products are as follows: 1. Smart Labeling (AI-assistant for data annotation) 2. Hyper-Scale Computing Framework 3. FAST-AI: Hyper-Scale Training (Automatic Parallel the AI model training task) 4. FAST-AI: Hyper-Scale Inference (Auto Scale the AI inference for more than 10000+ edge devices) 5. Full DGX End-to-End Deep Learning Solution 6. More edge hardware will be supported in 2021
The successful stories including to help the medical application client to speed up the AI training time for more than 3.5x faster compared to the state-of-the-art architecture. They have provided AI model training services for more than 15 enterprise-level manufacturing clients in Taiwan (They are in the vertical accelerator of NTUTEC and AUO in 2020).
In the next, they will launch the hyper-scale computing technique as AI-DevOps as a Service for global target users, the people who are doing deep learning programming. They can help them in the following tasks: 1. No installation for the AI development environment. 2. Automation speed-up the AI training speed without any coding. 3. Fast AI model deployment with their Hyper-scale computing inferencing.