I build systems that understand, generate, and translate speech—from model training and evaluation
to GPU inference and production deployment.
Multilingual ASR, large-scale data pipelines, and reliable evaluation.
TTS and streaming speech-to-speech translation.
Distributed training, low-latency GPU inference, benchmarking, deployment, observability, and CI/CD.
Production speech deployment.
ASR, TTS, and speech-to-speech services with immutable images, model hydration, scaling benchmarks, and automated releases.
Waki Demo Builder.
An agent/backend pipeline that turns meeting input into validated, sandboxed mini-app demos.
Python · PyTorch · Go · CUDA / GPU inference · Kubernetes · Docker · GitHub Actions
Apple · Siri TTS → Linktivity · Backend Systems → Kotoba Technologies · Speech AI
M.S. in Electrical and Computer Engineering, Carnegie Mellon University
Tokyo, Japan