Data Science & ML Claudeスキル

Channel Name Parsing Utility

Parses microscopy channel and marker names from KINTSUGI metadata files using automatic format detection.

Windows CuPy NVRTC Fix

Resolves CuPy runtime compilation errors on Windows by correctly configuring CUDA NVRTC paths.

Scientific Project Data Separation

Organizes scientific research repositories by decoupling core code from experimental data and notebook outputs.

HiPerGator Multi-Group Conda Manager

Configures and manages multiple Conda environment locations across different research group storage allocations on the UF HiPerGator supercomputer.

Jupyter Notebook Cell Referencing

Standardizes how AI identifies and communicates about specific Jupyter notebook cells using stable, identifiable characteristics instead of volatile cell numbers.

Cross-Modal Normalization for MaxFuse

Standardizes scale alignment for multi-modal RNA and protein data integration to ensure accurate cross-modal matching.

RL Trading Reward Optimizer v3.3.0

Optimizes reinforcement learning reward functions for automated trading to eliminate reward hacking and improve P&L gradient signals.

MaxFuse Bimodal Score Diagnosis

Diagnoses and resolves bimodal matching score distributions in MaxFuse multi-modal data integration pipelines.

Notebook Data Consistency

Prevents data shape mismatches and synchronization errors when modifying Jupyter notebook data sources.

Large Cell Ratio Matching for MaxFuse

Optimizes MaxFuse parameters to prevent RNA cell loss in datasets with extreme protein-to-RNA imbalance ratios.

Adaptive Predator-Prey Dynamics

Optimizes trading regime detection by dynamically adjusting Lotka-Volterra coefficients based on real-time market volatility and momentum.

Integrated Risk Manager

Implement unified position sizing and sophisticated risk management for trading algorithms using GARCH, Kelly criterion, and quadratic drawdown scaling.

Recursive Memory Explorer

Navigates complex memory graphs using RLM-style recursive search to find deep context without excessive token usage.

Ollama GPU Monitor

Monitors GPU performance, VRAM usage, and inference metrics for local Ollama models.

Ollama REST API Interface

Integrates Ollama's local LLM capabilities into Claude Code via direct REST API operations using the Python requests library.

LLM Application Evaluation

Evaluates Large Language Model application performance using automated metrics, human feedback loops, and LLM-as-judge frameworks.

OpenAI Compatibility for Ollama

Integrates Ollama with the OpenAI Python library to enable seamless migration and compatibility with OpenAI-based tools.

Ollama Python Integration

Integrates local LLM inference into Python applications using the official Ollama library for text generation, chat, and model management.

Prompt Engineering Patterns

Optimizes LLM performance and reliability through advanced prompt engineering techniques like few-shot learning and chain-of-thought.

HuggingFace Model Importer

Imports GGUF models directly from HuggingFace into Ollama for seamless local inference and experimentation.

ML Pipeline Workflow

Builds and orchestrates end-to-end MLOps pipelines from data preparation through model training, validation, and production deployment.