Machine Learning Intern - KWS/AED
Core
Build, evaluate, and improve deep learning models for keyword spotting (KWS) and audio event detection (AED) deployed on ultra-low-power edge hardware.
Role type
Machine Learning Intern (Audio/Speech)
Builds
Deep learning models for wake word detection, spoken commands, and acoustic event detection on NDP-class neural decision processors.
Domain
Edge AI, Audio Signal Processing, Semiconductor Hardware
Deliverable
production ML models
Required skills
Deep learning for audio/speech, Python, TensorFlow/Keras, audio signal processing fundamentals, ML evaluation concepts, model efficiency concepts
Preferred skills
Experience with CNNs/RNNs on spectrogram or time-series data, hard-negative mining, data augmentation strategies, model pruning/quantization
Technologies
Python, TensorFlow, Keras, PyTorch, CNN, RNN, Grad-CAM, PCEN, log-mel, filterbanks
Responsibilities
Support development and evaluation of KWS and AED models (single-stage and cascaded architectures), assist with audio pipeline and feature extraction, design and prune CNN architectures for hardware constraints, build false-accept diagnostic tooling, contribute to hard-negative mining and data augmentation, plan and track data collection efforts, analyze model run results, collaborate with ML/DSP/hardware engineers.
Seniority
Intern