3. Data, Science & AI

Provides end-to-end machine learning model training, evaluation, deployment, and MLOps workflow management.

Enables data analysis, statistical modeling, and visualization for reproducible research workflows in scientific and business contexts.

Efficient search utility for Toss/ Apps-in-Toss documentation using keyword and semantic similarity scoring with Korean-aware ranking.

Provides database design, schema optimization, indexing strategies, and migration management for efficient data storage and retrieval.

Guides data exploration, proposes visualization strategies, creates interactive Plotly charts in marimo notebooks, and generates analytical conclusions from datasets.

Guides multi-agent system design with decision trees, model comparison (Haiku vs Sonnet), and collaboration principles for agent selection.

Generates multiple LLM responses with probabilities to overcome mode collapse, enabling exploration of diverse answers for creative and complex tasks.

Transforms vague user requests into optimized, detailed prompts for AI coding agents, ensuring complete results in a single interaction.

Tracks user body weight entries and provides data aggregations, trend charts, and historical analysis for personal health monitoring.

Provides SQL best practices, including testing with pgTAP, linting via sqlfluff, query optimization, and migration management for data engineering workflows.

Processes and prepares genome assembly datasets for integration into VEuPathDB, a pathogen genomics database resource.

Manages database architecture, schema design, performance optimization, data governance, and migration for efficient data handling.

Provides lightning-fast data manipulation and analysis using a Rust-based DataFrame library, optimized for large datasets and ETL workflows.

Curates and processes bulk RNA-seq datasets for integration into VEuPathDB, enabling pathogen genomic analysis.

Guides structured creation and maintenance of CLAUDE.md memory files for hierarchical AI knowledge organization and team collaboration.

Expert guidance for designing Claude Code agents, covering architecture, prompt engineering, and tool selection best practices.

Manages enterprise user sessions and context with token budget optimization for efficient AI model interactions and state persistence.

Provides low-level orchestration for building stateful, durable AI agents using graph workflows, persistence, and human-in-the-loop capabilities.

Specializes in optimizing data formats including TOON, JSON, and YAML with serialization patterns and validation for modern application development.

Prioritizes local documentation and project patterns over external research, using AI (RAG, Perplexity) for necessary external queries.

Builds hybrid knowledge graphs by linking structured domain data (CSV) with lexical graphs from unstructured documents via LLMs, enabling GraphRAG and multi-source knowledge graph construction.

Fetches and parses Advent of Code puzzle data, extracting examples, expected outputs, and problem requirements using aoc-cli.

Provides optimized SQL patterns, indexing strategies, and EXPLAIN analysis to enhance database query performance and eliminate slow queries.

Provides persistent lifecycle management for AI agents, enabling cross-turn survival, workflow-scoped execution, and automatic cleanup.