I am an Embedded Design Engineer who enjoys making hardware listen, think, and talk. I specialize in bridging the gap between bare-metal silicon and heavy AI models, with a specific focus on Audio and Edge AI.
If it compiles, ships, and fits in 256KB RAM, I'm happy.
- The Metal: C, C++, Rust, ESP32, RISC-V
- The OS: Embedded Linux, Yocto, Zephyr RTOS, ALSA/ASoC
- The Brains: TensorFlow Lite, PyTorch, Edge Impulse, NPU Acceleration
- The Tools: Neovim, Docker, Git, KiCad, Logic Analyzers
- Low-Compute ASR: Building a Speech Recognition model using Convolutional Vision Transformers (CvT) + CTC, specifically designed for low-power targets.
- Embedded Audio: Writing Linux kernel drivers for Dante networked audio and offloading DSP tasks.
- Hardware: Typing this on a custom ZMK split keyboard (Seed Xiao ESP32-C6) that I built and soldered myself.
- Whisper on the Edge: Optimized a C++ pipeline to run Whisper/Wav2Letter on NPUs, cutting latency down to ~5s for 30s audio.
- Real-time Vision: Built a dashcam-based littering detection system using OpenCV and Edge AI.
- Yocto Wrangler: I actually enjoy writing
.bbappendfiles and debugging Device Trees.
- LinkedIn: linkedin.com/in/snappiersoap
- Email: official.adwaith.s@gmail.com