Leanpub Drops Edge AI Android Kotlin Book

This is the kind of AI book that matters once you move past demos: the hard problems are thermals, memory bandwidth, and heterogeneous accelerators, not model slogans. The NNAPI and AICore angle makes it relevant for teams trying to ship across fragmented Android hardware, where portability is part of the performance problem. Quantization, pruning, and QAT are the right levers for on-device workloads, especially when 60 FPS and battery life are both non-negotiable. Kotlin 2.x plus zero-copy NDK pipelines suggests a production-minded guide, not a generic ML primer. The strongest audience is mobile teams building camera, audio, or always-on assistant features that need predictable latency and thermal stability. It sits squarely in the edge AI niche, but it is practical developer news rather than theory or research.