Own the audio front-end pipeline for Dyno — the full path from microphone capture to clean signal delivered into downstream ASR, wake-word, and conversational AI.
Design, implement, and optimize core DSP algorithms: acoustic echo cancellation (AEC), noise suppression (NS), dereverberation, and adaptive filtering.
Solve open, hardware-specific problems with no off-the-shelf solution — e.g. suppressing the robot's own ego-noise (motors, fans, joint actuators) during live capture.
Set the architecture and quality bar for the entire audio front-end: define signal-quality targets, latency budgets, and evaluation methodology.
Ensure all processing runs real-time and low-latency on embedded / edge hardware within tight compute and power constraints.
Partner with hardware and mechanical teams on microphone selection, array layout, and acoustic calibration of the robot chassis.
Establish measurement, tuning, and regression frameworks so audio quality is reproducible and does not regress across builds.
Act as the technical authority on audio for the organization; mentor engineers and guide technology direction without needing a management headcount.
REQUIREMENTS
Experience: 8+ years in audio signal processing / acoustic engineering, with demonstrated depth in far-field speech capture.
Core domain: Deep expertise in AEC, noise suppression, dereverberation, and adaptive filtering — down to the underlying math, not just library usage.
Signal processing: Strong DSP fundamentals — adaptive filters (NLMS/RLS/Kalman), spectral processing, statistical signal processing, psychoacoustics.
Programming: Expert C/C++ for real-time embedded audio; Python for prototyping and analysis.
