3. Data, Science & AI

Orchestrates CSV data imports and list-item processing for the raw data lake feeding the LUCI system.

Central AI orchestrator managing task routing, fallbacks, caching, and usage tracking for efficient AI workflow execution.

Command-line interface for interacting with Model Context Protocol (MCP) servers, enabling access to AI tools, APIs, and data sources via terminal commands.

Queries token usage, costs, and session statistics for Claude Code and Codex AI models via a command-line interface.

Automates specialist agent selection by intelligently analyzing user requests without explicit mentions, optimizing task routing.

Reviews Custom Agent quality and provides improvement suggestions for AI agent development and optimization.

Generates role-specific AI sub-agents for Claude Code, enabling delegated task execution with permission separation.

Expert guidance for building AI agents via Claude Agent SDK, covering architecture, tool design, and MCP server integration for production deployment.

Updates an AI agent's working memory via background daemon for non-blocking, single-operation memory management.

Automates Zendesk API interactions to retrieve, test connections, and export ticket data for streamlined data management.

Publishes and manages AI research papers on Hugging Face Hub, enabling model/dataset linking and professional markdown article generation.

Converts Product Requirement Documents (PRDs) into the structured JSON format required by the Ralph autonomous agent system.

Solves constraint satisfaction problems including Sudoku, N-queens, and scheduling using constraint propagation and search algorithms.

Optimizes constraint satisfaction problem solving by propagating constraints to eliminate impossibilities and fail fast on contradictions before search.

Builds abstract syntax trees for symbolic mathematical expressions, enabling code manipulation and transformation in computational mathematics workflows.

Solves NP-hard optimization problems (TSP, scheduling) using greedy construction and local improvement techniques (2-opt, hill climbing).

Manages local Ollama LLM models for development and testing, including configuration, model switching, and OpenAI-compatible API integration.

Formats statistical results into aligned, human-readable tables with summary statistics for clear data reporting.

Generates spanning trees and random mazes via frontier-based exploration with configurable traversal order for graph coverage tasks.

Calculates permutations, combinations, and probability distributions using Monte Carlo simulations and brute-force enumeration for card/dice problems.

Provides utilities for counting element frequencies, generating histograms, and analyzing data distributions.

Computational geometry tool for finding convex hulls, point enclosure, and polygon operations using the monotone chain algorithm.

Verifies solution correctness independently of computation method, ensuring accuracy in data and AI result validation.

Enables semantic search across ASU GitHub repositories using RAG to discover code patterns, integrations, and ASU-specific conventions before integration tasks.