Build comprehensive ML pipelines, experiment tracking, and model registries with MLflow, Kubeflow, and modern MLOps tools. Implements automated training, deployment, and monitoring across cloud platforms. Use PROACTIVELY for ML infrastructure, experiment management, or pipeline automation.
Optimize vector index performance for latency, recall, and memory. Use when tuning HNSW parameters, selecting quantization strategies, or scaling vector search infrastructure.
Master Rust 1.75+ with modern async patterns, advanced type system features, and production-ready systems programming. Expert in the latest Rust ecosystem including Tokio, axum, and cutting-edge crates. Use PROACTIVELY for Rust development, performance optimization, or systems programming.
Use when working with object manipulation in Stacks — deep merging with type safety, object mapping/transformation, strict key checking, typed entries/keys, property picking, clearing undefined values, or the DeepMerge utility type. Covers @stacksjs/objects.
End-to-end autonomous pipeline that runs auto-review-fix, then auto-pr-merge
Machine learning pipeline for scientific research including data preprocessing, feature engineering, model selection, training, evaluation, and interpretation. Covers supervised/unsupervised learning, deep learning, cross-validation, hyperparameter tuning, and model explainability. Use when user asks to build a predictive model, classify data, cluster samples, do feature selection, or apply ML to research data. Triggers on "machine learning", "classification", "clustering", "random forest", "neural network", "deep learning", "predict", "feature selection", "cross-validation", "train model".
Machine learning with scikit-learn. Use when: classification, regression, clustering, dimensionality reduction, model evaluation, feature engineering. NOT for: deep learning (use transformers/pytorch), time series forecasting (use statsmodels), big data (use spark).
HuggingFace Transformers for model inference. Use when: text classification, NER, question answering, summarization, embeddings, zero-shot classification. NOT for: training large models (use cloud), simple regex/rule-based tasks, production serving at scale (use vLLM).
Track Clawdbot AI model usage and estimate costs. Use when reporting daily/weekly costs, analyzing token usage across sessions, or monitoring AI spending. Supports Claude (opus/sonnet), GPT, and Codex models.
Monitor Minimax Coding Plan usage to stay within API limits. Fetches current usage stats and provides status alerts.
Orchestrate multiple AI models (GLM, MiniMax, etc.) as workers using Pi Coding Agent with Claude as coordinator.
A comprehensive AI model routing system that automatically selects the optimal model for any task. Set up multiple AI providers (Anthropic, OpenAI, Gemini, Moonshot, Z.ai, GLM) with secure API key storage, then route tasks to the best model based on task type, complexity, and cost optimization. Includes interactive setup wizard, task classification, and cost-effective delegation patterns. Use when you need "use X model for this", "switch model", "optimal model", "which model should I use", or to balance quality vs cost across multiple AI providers.
Create a modular typography scale with size, weight, and line-height relationships.
Guide for adding support for new LLM or VLM models in Megatron-Bridge. Covers bridge, provider, recipe, tests, docs, and examples. Use when the user asks to add, support, onboard, or integrate a new model, or when creating bridges, providers, or recipes for a new model family.
Run Megatron-LM (MLM) and Megatron Bridge training with mock or real data. Covers correlation testing, available recipes, and multi-GPU examples. Use when running training, comparing MLM vs Bridge, or translating configs.
Structured framework for verifying numerical parity of HF<->MCore weight conversions. References existing tools and the add-model-support skill. Use when debugging weight mismatches, verifying checkpoint round-trips, or choosing which verification tool to run.
Build DAG-based AI pipelines connecting Gradio Spaces, HuggingFace models, and Python functions into visual workflows. Use when asked to create a workflow, build a pipeline, connect AI models, chain Gradio Spaces, create a daggr app, build multi-step AI applications, or orchestrate ML models. Triggers on: "build a workflow", "create a pipeline", "connect models", "daggr", "chain Spaces", "AI pipeline".
Project knowledge accumulation system - learn from practice, avoid repeating mistakes
Pattern extraction, confidence-scored evaluation, skill creation, organization, versioning, and cross-project export pipeline.
Complexity-based task routing with Q-Learning optimization, Agent Booster WASM fast-path, and Mixture-of-Experts model selection.
Few-shot example generation and optimization for improved LLM performance
Hugging Face transformer model fine-tuning and inference for intent classification
