Data Science & ML Claudeスキル

Speech-to-Text

Implements high-performance, privacy-focused speech recognition systems using Faster Whisper for secure audio transcription and voice assistant integration.

Text-to-Speech

Implements high-performance, secure text-to-speech synthesis using Kokoro TTS for real-time voice applications.

Funsloth RunPod Training Manager

Automates the lifecycle of LLM fine-tuning workloads on RunPod GPU instances using the Unsloth framework.

Hugging Face Training Manager

Simplifies the deployment and management of Unsloth fine-tuning jobs on Hugging Face cloud GPUs.

Model Quantization

Optimizes Large Language Models through 4-bit and 8-bit quantization to enable efficient deployment on resource-constrained hardware.

Funsloth Dataset Validator

Validates and analyzes datasets to ensure they are optimized for Unsloth fine-tuning and Chinchilla optimality.

Funsloth Model Upload

Automates the generation of professional model cards and the deployment of fine-tuned LLMs to the Hugging Face Hub.

FunSloth Unsloth Training Generator

Generates optimized training notebooks and scripts for fine-tuning LLMs using the Unsloth framework.

Funsloth Local GPU Training

Manages local GPU fine-tuning workflows using Unsloth to optimize LLM training performance and resource utilization.

Senior Data Scientist

Provides expert guidance and automated tools for statistical modeling, experiment design, causal inference, and production-grade machine learning pipelines.

Wake Word Detection

Implements high-performance, privacy-preserving wake word detection for always-listening voice assistants using openWakeWord and test-driven development.

Financial Modeling Suite

Builds advanced financial models including DCF analysis, Monte Carlo simulations, and sensitivity testing for robust investment decision-making.

BUSCO Phylogenomics Workflow Generator

Generates comprehensive, scheduler-aware phylogenomics pipelines from genome assemblies using single-copy orthologs.

BAML Code Generation

Generates production-ready BAML applications for structured data extraction, classification, and AI agent workflows using real-time patterns from official BoundaryML documentation.

Prompt Engineering Toolkit

Systematically designs and optimizes Claude prompts using a multi-level methodology for maximum effectiveness and cost efficiency.

ElevenLabs Text-to-Speech

Converts text into high-quality audio files using the ElevenLabs API with customizable voice parameters and presets.

BAML Code Generation

Generates production-ready BAML applications and structured output configurations from natural language requirements.

LLM Provider Management

Manages multi-provider patterns and raw SDK access for OpenAI, Anthropic, Google Gemini, and Ollama.

Data Visualization Techniques

Transforms complex datasets into clear, actionable visual narratives through optimized chart selection, dashboard design principles, and modern React charting libraries.

Agent Creator

Designs and builds custom agents, skill frameworks, and Model Context Protocol (MCP) integrations using specialized personas and expert knowledge patterns.

Flox CUDA Development

Streamlines NVIDIA CUDA setup and GPU computing environments using Flox for deep learning and high-performance development.

AI Integration Patterns

Implements production-ready AI capabilities including prompt engineering, streaming responses, and cost-effective context management.

Episode Start

Initializes tracking sessions for AI execution patterns to enable continuous self-learning and behavioral evolution within agentic workflows.

Research Superpowers Getting Started

Orchestrates systematic literature searches and reviews by finding, screening, and extracting data from scientific papers across PubMed and Semantic Scholar.

Experiment Design Methodology

Standardizes the creation, execution, and analysis of A/B tests and product experiments to ensure statistical significance and actionable insights.

BioGeoBEARS Biogeographic Analysis

Automates the setup, validation, and execution of phylogenetic ancestral range reconstruction using BioGeoBEARS in R.

Multi-Agent System Architecture

Designs and implements sophisticated multi-agent workflows using modular orchestration patterns, hierarchical delegation, and deterministic tool coordination.

Randomization

Implements rigorous random assignment procedures for scientific experiments to minimize selection bias and meet CONSORT standards.

Blinding Procedures

Minimizes experimental bias by implementing structured blinding protocols and objectivity standards for scientific research studies.

AI Multimodal Processing

Processes and generates audio, video, images, and documents using the Google Gemini API to enable advanced media understanding and creation within Claude.