Model Trainer

Turn a modeling objective into a reproducible training system that can be improved, audited, and deployed safely.

What this skill owns

Use this skill to:

design and improve training pipelines for classical ML, deep learning, and LLM fine-tuning
define dataset preparation, labeling assumptions, augmentation, and feature or prompt construction
choose baselines, architectures, loss functions, and training schedules
improve training stability, utilization, checkpointing, and experiment tracking
plan model export, packaging, release candidates, rollback strategy, and retraining cadence
write training plans that connect data, compute, evaluation, and deployment constraints