ML Master

This skill should be used when the user asks to productionize, track, version, govern, monitor, or automate ML systems. PROACTIVELY activate for: (1) MLflow, Weights & Biases, Neptune, Comet, ClearML experiment tracking, (2) model registry, model versioning, artifact lineage, reproducibility, (3) Kubeflow, SageMaker Pipelines, Vertex AI Pipelines, Azure ML pipelines, Databricks workflows, (4) CI/CD, continuous training/evaluation, A/B tests, canary/shadow deployments, (5) drift detection, model monitoring, data validation, responsible AI governance. Provides: end-to-end MLOps architecture and operational safeguards.

This skill should be used when the user asks to train, debug, scale, or improve ML models. PROACTIVELY activate for: (1) PyTorch, TensorFlow/Keras, JAX, Flax, Hugging Face Trainer/Accelerate training loops, (2) distributed training, DDP/FSDP/DeepSpeed, TPU/GPU setup, (3) mixed precision AMP/bf16, gradient accumulation, checkpointing, seeding, (4) overfitting, imbalance, loss functions, regularization, LR schedules, warmup, (5) memory optimization, gradient checkpointing, offloading, quantization-aware training. Provides: reproducible training best practices across deep learning and classical ML.

This skill should be used when the user asks to evaluate, validate, compare, explain, or debug model performance. PROACTIVELY activate for: (1) metrics selection for classification, regression, ranking, NLP, CV, recommender, forecasting, and generative AI, (2) train/validation/test splits, cross-validation, grouped or time-series validation, (3) confusion matrices, ROC/PR curves, calibration, thresholds, error analysis, (4) ablation studies, statistical significance, confidence intervals, bootstrap tests, (5) bias, fairness, explainability, robustness, leakage detection. Provides: rigorous evaluation methodology and production-readiness checks.

This skill should be used when the user asks to adapt pretrained, foundation, language, vision, multimodal, or embedding models. PROACTIVELY activate for: (1) transfer learning, full fine-tuning, frozen backbones, adapters, LoRA, QLoRA, AdaLoRA, PEFT, (2) Hugging Face Transformers, Diffusers, Accelerate, TRL, RLHF, DPO, preference tuning, alignment, (3) dataset preparation, instruction tuning, chat templates, tokenization, packing, catastrophic forgetting, (4) RAG, vector databases, embedding optimization, reranking, (5) multimodal and edge fine-tuning. Provides: safe, efficient fine-tuning and adaptation guidance.

This skill should be used when the user asks to ingest, clean, validate, transform, version, monitor, or serve ML data and features. PROACTIVELY activate for: (1) data ingestion, preprocessing, feature engineering, leakage prevention, train/serving skew, (2) Spark, Dask, Polars, pandas, Ray Data, streaming pipelines, (3) Great Expectations, TFDV, Deequ, data quality and validation, (4) DVC, lakehouse tables, dataset versioning, lineage, reproducibility, (5) Feast, Tecton, Hopsworks feature stores, point-in-time joins, online/offline features. Provides: scalable, reproducible, leakage-safe ML data pipeline design.

This skill should be used when the user asks to deploy, scale, or cost-optimize ML workloads on cloud platforms. PROACTIVELY activate for: (1) AWS SageMaker Studio, Training, Processing, Pipelines, Endpoints, Model Monitor, Feature Store, Clarify, Ground Truth, EC2 GPUs, EKS, Lambda, Inferentia, Trainium, (2) GCP Vertex AI Training, Pipelines, Endpoints, Feature Store, Model Monitoring, AutoML, Matching Engine, TPU, GKE, Cloud Run, (3) Azure ML workspaces, pipelines, managed endpoints, AutoML, Responsible ML, AKS/ACI, (4) Databricks, Modal, Replicate, RunPod, Lambda Labs, Anyscale. Provides: cloud ML architecture, autoscaling, hardware, security, and cost guidance.

This skill should be used when the user asks to tune hyperparameters, run sweeps, optimize search spaces, or use AutoML. PROACTIVELY activate for: (1) Optuna, Ray Tune, FLAML, AutoGluon, Hyperopt, Nevergrad, KerasTuner, W&B sweeps, (2) grid search, random search, Bayesian optimization, TPE, Gaussian processes, evolutionary search, (3) ASHA, Hyperband, successive halving, multi-fidelity optimization, population-based training, (4) learning-rate finder, batch-size search, early stopping, pruning, (5) reproducible sweep design and experiment analysis. Provides: budget-aware hyperparameter search strategy.

This skill should be used when the user asks to automate Azure ML and Azure Data Factory production workflows. PROACTIVELY activate for: (1) Azure ML code asset registration, azure-ai-ml SDK, AML code versions, `result.version`, requested-vs-actual versions, (2) ADF to Azure ML orchestration, ADF WebActivity, managed identity blob reads, `connectVia`, managed VNet IR, (3) code asset version pointer blobs, latest.json contracts, training/scoring code version propagation, (4) private storage firewalls, Microsoft-hosted CI agents, temporary network rules, storage data-plane smoke tests, (5) marshmallow<4 pinning, AML SDK import failures, runtime validation for Azure ML infrastructure. Provides: operationally safe Azure ML + ADF automation patterns.

This skill should be used when the user asks to optimize, export, serve, compress, or accelerate ML inference. PROACTIVELY activate for: (1) latency, throughput, p95/p99, batching, concurrency, KV cache, memory, or cost issues, (2) quantization INT8/INT4, GPTQ, AWQ, bitsandbytes, pruning, sparsity, distillation, (3) ONNX export, ONNX Runtime, TensorRT, TorchScript, torch.compile, XLA, OpenVINO, Core ML, TFLite, (4) Triton, TorchServe, TF Serving, BentoML, Seldon, KServe configuration, (5) edge deployment, CPU/GPU/TPU/Inferentia serving. Provides: hardware-aware inference optimization and safe benchmarking.