Posts tagged with "machine-learning"

The standard SDLC model breaks for data science because model training requires production data. Here is a five-stage promotion framework that resolves the conflict between data scientist flexibility and production security controls.

Exploring whether modern PyTorch ASR pipelines expose phoneme-level representations, using Wav2Vec 2.0 to extract and visualize phoneme probabilities from speech.

Resurrecting a technical blog after seven years away — migrating old WordPress archives, Google Blogger posts, and 1990s DarkMagic.org content into a new Jekyll site on GitHub Pages.