Finding signal in the noise.
鼓動・遊び・対比
I build deep-learning systems for signal processing at scale — pulling robust structure out of noisy, high-volume data, and shipping it so it holds up in production.
Selected work仕事
A few things I’ve built — signal analysis at clinical scale, learned reconstruction, and interpretable models.
Approach手法
Four threads run through everything I build — from a clinical ECG to a survey of the sky, the problem rhymes.
Signal detection in noise信号検出
Pulling structure out of noisy, high-volume signals — arrhythmia and anomaly detection, denoising, time-frequency analysis, signal-quality assessment. The same detection-in-noise problem astronomy faces with transients and faint sources.
Deep learning at scale大規模学習
Training and shipping models against millions of records: data pipelines, HPC and GPU compute, and production ML that stays reliable far past the prototype.
Explainable & trustworthy AI説明可能性
Making model decisions legible — SmoothGrad, Grad-CAM, integrated gradients — so predictions can be trusted in clinical and scientific settings.
Research engineering & systems研究基盤
From Linux-kernel and embedded systems to modern ML infrastructure: reproducible, well-tested software, not just notebooks — with a bias toward correctness.
Contact連絡
Open to interesting problems in signal, scale, and learning. The fastest way to reach me is email.
Open to
- Research & engineering roles in signal / ML at scale
- Collaborations on time-series and detection problems
- Speaking, peer review, and technical writing