Dear QQ ,what can I do to reach your technical level?

[yuyun2000]

Are you a student? Or have you been working for many years? What do you think about the prospects of audio model deployment?

[DakeQQ]

Hello,

Thank you for your interest in my technical background.

Since 2022, we've been working in AI, focusing on deploying models on edge devices. Audio model deployment is particularly exciting, with applications ranging from voice assistants to accessibility tools. As demand grows for low-energy, cost-effective solutions, investing in this field is increasingly attractive.

Successful deployment goes beyond models, requiring expertise in hardware challenges, model algorithms, operator optimization, and firmware. Mastering these areas is time-intensive but offers rewarding career prospects.

For those interested in edge AI, consider these frameworks:

1. ONNX Runtime: Great operator support, compatible with Qualcomm QNN, though limited in CPU-FP16 and INT4 performance.
2. LiteRT (TFLite): Supports Qualcomm QNN and GPU/NPU tasks, but lags slightly in PyTorch model operator conversion.
3. MNN: Offers top CPU performance for FP16/INT4, but lacks Qualcomm NPU support.

These frameworks open doors to lucrative job opportunities. By exploring their APIs and testing models, you'll master the deployment pipeline.

Wishing us both success in edge AI.

Best regards

[yuyun2000]

Salute to you!