Data Science & ML Claude 技能

Flow Nexus Neural Networks

Trains and deploys complex neural network architectures within distributed E2B sandbox environments for scalable machine learning workflows.

Prompt Engineering Lab

Creates, optimizes, and debugs high-performing, production-ready prompts for Claude 4, GLM 4.7, and Gemini 3 using evidence-based techniques.

Scikit-learn Machine Learning

Builds, evaluates, and deploys production-ready machine learning models using the industry-standard scikit-learn library.

AgentDB Learning Plugins

Implements and trains advanced reinforcement learning algorithms to create autonomous agents that evolve through experience.

AI Gateway Manager

Manages and routes requests across multiple AI providers including Anthropic, Ollama, and HuggingFace.

Structural Biology Analysis

Analyzes protein and molecular structures through AlphaFold interpretation, quality validation metrics, and comparative structural techniques.

RAG Implementation

Implements Retrieval-Augmented Generation (RAG) systems with vector databases and semantic search to build grounded, knowledge-aware AI applications.

Hybrid Search Implementation

Combines vector similarity and keyword-based search to optimize retrieval accuracy in RAG systems and search engines.

Similarity Search Patterns

Implements high-performance similarity search and vector database patterns for semantic retrieval and RAG systems.

PyTorch Geometric (PyG)

Develops and trains Graph Neural Networks (GNNs) for node classification, link prediction, and geometric deep learning tasks.

Prompt Engineering Patterns

Optimizes Large Language Model performance through advanced reasoning patterns, few-shot learning, and structured prompt templates.

Risk Metrics Calculation

Calculates comprehensive portfolio risk metrics and performance indicators for quantitative trading strategies and investment management.

HPC Development & Testing Workflow

Streamlines scientific development on HPC environments using multi-root workspaces and automated test data extraction.

Quantitative Backtesting Frameworks

Builds production-grade backtesting systems for trading strategies while mitigating common biases and accounting for realistic market costs.

LangChain Architecture Design

Builds sophisticated LLM applications using the LangChain framework to implement autonomous agents, complex workflows, and persistent memory.

Vector Index Optimization

Optimizes vector database indexes for production performance by balancing search latency, recall accuracy, and memory consumption.

Plotly Interactive Visualization

Generates interactive, publication-quality Python charts and dashboards for data exploration and presentation.

GPU Parallel Scheduling & Memory Safety

Implements queue-based GPU allocation and memory cleanup patterns to prevent OOM crashes and ensure reliable progress tracking in parallel workflows.

RAG Implementation Expert

Builds production-grade Retrieval-Augmented Generation (RAG) systems using vector databases and semantic search for LLM applications.

MLOps Pipeline Orchestrator

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

Embedding Strategies for RAG

Optimizes embedding model selection and chunking strategies to improve semantic search and RAG application accuracy.

Data Storytelling

Transforms complex datasets into persuasive narratives and executive-ready presentations using proven storytelling frameworks.

LLM Evaluation Framework

Implements comprehensive evaluation strategies for LLM applications using automated metrics, human feedback, and benchmarking.

Airflow DAG Patterns

Builds production-ready Apache Airflow DAGs using industry-standard patterns for orchestration and data engineering.

Multimodal Media Analysis

Analyzes and extracts deep insights from images, videos, and audio files using advanced AI models.

DuckDB & MotherDuck Analytics

Streamlines building high-performance OLAP applications using DuckDB, MotherDuck, and Parquet in Node.js and TypeScript.

Scientific Exploratory Data Analysis

Performs automated exploratory data analysis and generates comprehensive reports for over 200 scientific file formats.

Genomics & Transcriptomics Analysis

Provides strategies, implementation patterns, and workflows for genomics and transcriptomics data analysis.

OpenRouter Multi-Model Gateway

Integrates 300+ AI models into Claude Code for specialized tasks, high-fidelity image generation, and cross-model reasoning.

LangChain Architecture

Builds sophisticated LLM applications and autonomous agents using the LangChain framework's core patterns and integrations.