3. Data, Science & AI

Builds type-safe conversational AI agents in Python using Pydantic AI and OpenRouter, supporting tool calling, validation, and streaming.

CLI tool for validating Model Context Protocol (MCP) server implementations in AI model development workflows.

Analyzes variance and discrepancies in friction pipeline outputs across runs, thinking levels, and seeds for visa research and route selection consistency.

Analyzes user-uploaded screenshots to extract visual content, identify elements, and describe context from images in the Pictures directory.

Analyzes Polymarket prediction markets to identify mispriced events and arbitrage opportunities through quantitative data analysis.

Command-line interface for BigQuery operations including data querying, table management, and cost estimation via dry-run.

Executes SQL queries on AWS Athena against telemetry data stored in Parquet format, supporting Glue catalog exploration and partition debugging.

Multi-format document parsing tools for text, table, and structure extraction from PDF, DOCX, HTML, and Markdown, supporting RAG pipeline development.

Orchestrates multiple AI agents in parallel for independent tasks, handling decomposition, agent selection, and conflict prevention without shared state.

Automates addition of missing Octopus card transactions from app history, enriches with duration metadata, and records card reloads.

Generates and edits images using Google's Nanobanana AI model via Replicate API, supporting text-to-image, image-to-image, and style transfer tasks.

Specializes in building AI-driven recommendation systems using Google AI SDK to deliver personalized content suggestions.

Provides real-time voice AI capabilities including STT, TTS, and LiveKit integration for voice agent pipelines.

Analyzes D1 database queries for performance optimization and error prevention using static analysis techniques.

Framework for analyzing Claude's configuration, identifying knowledge gaps, and proposing skill improvements to optimize AI assistance capabilities.

Provides domain expertise for sleep science data, including hypnograms, sleep stages, and metrics in backend sleep tracking systems.

Enables comprehensive spreadsheet creation, editing, and analysis with formula handling, formatting, and data visualization across multiple spreadsheet formats.

Builds a streaming AI chatbot with Groq API, RAG context, and Next.js, featuring floating UI and Vercel deployment.

Converts PDF, Word, images (with OCR), audio (with transcription), and other formats to Markdown for text analysis and LLM processing.

Automatically creates standardized Model Card reports in Markdown from data analysis and machine learning notebook results.

Professional data analysis workflow using Python ecosystem tools to extract deep insights from datasets.

Enables creation and configuration of custom subagents in Claude Code with domain-specific prompts, tool restrictions, and validation rules for specialized AI workflows.

Provides patterns for building AI conversational agents, including personas, insight pipelines, and Slack/Kakao integrations.

Designs and optimizes high-performance prompts using Anthropic's best practices to enhance AI model interactions.