Data Science & ML Claude 技能

Multi-Timeframe RL Training

Trains Reinforcement Learning models across multiple market timeframes with automated data resampling and professional market alignment.

Trading Backtest Capital Accounting

Corrects financial logic in trading simulators to prevent inflated equity curves and inaccurate drawdown metrics.

Resampling Strategies for Tidy Models

Implements robust model validation and resampling techniques using the R tidymodels ecosystem and rsample package.

Training Code Verification

Validates machine learning training scripts and configurations to ensure path accuracy, weight consistency, and environment compatibility.

Model Version Protocol

Manages version compatibility between trained trading models and live trading environments by embedding metadata into checkpoints.

MaxFuse Notebook Import Conventions

Standardizes import patterns and file paths for MaxFuse Jupyter notebooks to ensure reliable package integration.

Quarto Anti-Slop

Enforces reproducible research and academic writing standards for Quarto and RMarkdown documents to eliminate generic AI-generated content.

Kalshi Prediction Markets

Accesses real-time prediction market data, betting odds, and trading analytics from the Kalshi platform.

Trading Symbol Database Selection

Optimizes financial market universe selection using a multi-stage pipeline and SQLite-backed data management.

Reward Scaling Calibration

Calibrates reinforcement learning reward scales to eliminate unrealistic drawdown metrics in algorithmic trading models.

Trading Model Post-Training Workflow

Validates and assesses reinforcement learning trading models through systematic gating, backtesting, and walk-forward validation.

Stitched Image Quality Control

Automates quality control for stitched microscopy images by detecting saturation failures and visible tile grid patterns.

GPU-Aware ML Training Configuration

Optimizes PPO training performance on A100 and H100 GPUs by automatically aligning hyperparameters with hardware capabilities.

Superpowers Lab

Provides a specialized laboratory environment for experimenting with and implementing advanced Claude capabilities.

LLM Integration & Orchestration

Integrates multiple LLM providers like Anthropic, OpenAI, and Google Gemini into applications using advanced orchestration and reasoning patterns.

Machine Learning Pipeline Workflow

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

Trading Strategy Backtester

Backtests and evaluates quantitative trading strategies using historical stock market data to generate performance metrics and visualizations.

AgentDB Reinforcement Learning

Implements and trains autonomous agents using nine specialized reinforcement learning algorithms for self-improving behavior.

Recommendation Engine Systems

Constructs personalized recommendation engines using collaborative and content-based filtering to deliver tailored user experiences.

Flow Nexus Neural

Train, deploy, and manage distributed neural networks across sandboxed E2B environments using multiple architectures and federated learning.

Ollama Model Management & Optimization

Optimizes local LLM orchestration and GPU performance for Ollama-integrated AI environments.

TextCleaner Pipeline Manager

Manages and automates complex text transformation pipelines via the TextCleaner REPL interface.

Vector Databases

Guides developers through selecting, implementing, and optimizing vector search solutions for AI applications and RAG pipelines.

Technical Stock Indicators

Calculates core financial technical indicators including MA, MACD, RSI, and Bollinger Bands for stock market analysis and quantitative trading.

Market Sizing Analysis

Calculates TAM, SAM, and SOM using top-down, bottom-up, and value theory methodologies to quantify business opportunities.

LLM Training & Finetuning

Streamlines the process of training and finetuning large language models using industry-standard frameworks and memory optimization techniques.

ML Model Deployment Helper

Automates the productionization and deployment of machine learning models into scalable environments using FastAPI, Docker, and Kubernetes.

Docling Advanced Document Extraction

Extracts structured data from complex PDFs, scanned documents, and multi-column layouts using the advanced Docling engine and Granite vision-language models.

Classification Model Builder

Automates the end-to-end creation, evaluation, and deployment of machine learning classification models from natural language requests.

AgentDB Memory Patterns

Implements persistent, high-performance memory and learning patterns for AI agents using AgentDB.