Data Science & ML Claude 스킬

Grail Miner Management

Optimizes the deployment, operation, and performance of Grail miners on Bittensor Subnet 81 for verifiable language model post-training.

Ollama Client

Interfaces with the Ollama API to perform high-performance text completions and clinical analysis using the Phi-4 language model.

Tile Snake Pattern Stitching

Reconstructs multiplex microscopy images by correctly ordering tiles acquired via snake or serpentine stage patterns.

Data Ingestion Builder

Automates the creation of robust data ingestion providers using standardized patterns for registry mapping, storage abstraction, and metadata tracking.

Kalshi Prediction Markets

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

Markov Trading Regime Features

Implements and debugs dynamic Hidden Markov Model (HMM) regime detection for Reinforcement Learning trading observations.

Prompt Engineering Patterns

Optimizes LLM performance and reliability through advanced techniques like few-shot learning, chain-of-thought reasoning, and modular prompt templates.

Hugging Face Spaces Manager

Manages and interacts with Hugging Face Spaces by providing tools to search repositories, retrieve metadata, and control application runtimes.

LLM Evaluation Framework

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

RAG Implementation for LLM Apps

Builds high-performance Retrieval-Augmented Generation systems using vector databases and semantic search to ground AI responses in external knowledge.

LangChain Architecture & Agent Design

Architects sophisticated LLM applications using LangChain agents, multi-step workflows, and advanced memory management systems.

Incentive-Based Prompting

Optimizes Claude's output quality using research-backed psychological framing and statistical pattern-matching techniques.

Agent Builder

Orchestrates the development of complete, production-ready Claude Code agents using Anthropic's official best practices for context engineering and tool design.

LA-Bench Procedure Generator

Generates detailed experimental procedures from LA-Bench datasets by orchestrating a multi-agent workflow for parsing, reference fetching, and quality validation.

Machine Learning Pipeline Workflow

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

Health Economics Analysis in R

Performs advanced health economic evaluations including cost-effectiveness analysis, Markov modeling, and probabilistic sensitivity analysis in R.

RDKit Cheminformatics

Performs advanced molecular modeling and cheminformatics tasks including descriptor calculation, fingerprinting, and substructure searching.

Analyze Goodreads Export

Queries and analyzes personal book libraries from Goodreads CSV exports to provide reading insights and statistics.

LangGraph Fine-Tuning

Optimizes LangGraph application performance by iteratively refining prompts and node-level processing logic based on quantitative evaluation criteria.

ML API Endpoint Expert

Streamlines the design and deployment of production-grade machine learning inference endpoints using FastAPI and Docker.

Gemini File Search Tool

Manages Google Gemini File Search stores and implements production-ready RAG systems for documents and codebases.

LLM CLI

Interacts with diverse large language models through a command-line interface to perform tasks like prompt execution, data extraction, and embedding management.

Behavior Tree Composer

Generates and validates executable Python behavior trees for robotic systems using natural language task descriptions.

Parakeet Audio Transcription

Converts audio files into high-quality timestamped transcriptions using NVIDIA's Parakeet model optimized for Apple Silicon.

LA-Bench Workflow

Orchestrates the end-to-end generation, validation, and iterative improvement of detailed laboratory experimental procedures using the LA-Bench framework.

Innovation Game Miner

Develops, optimizes, and submits high-performance competitive algorithms for Boolean Satisfiability, Vehicle Routing, and Knapsack challenges on The Innovation Game platform.

Data Cleaner

Automates dataset analysis and cleaning by detecting data types, identifying quality issues, and generating Python scripts for standardized data preparation.

Document Ranking and Scoring

Ranks and filters retrieved documents based on vector similarity metrics to optimize RAG pipeline relevance.

Polars Data Processing

Provides expert guidance for high-performance data manipulation and ETL pipeline optimization using the Polars library in Python.

TTS LiveKit Plugin

Deploys a self-hosted Text-to-Speech API using MeloTTS and integrates it with LiveKit voice agents for high-quality, low-latency speech synthesis.