3. Data, Science & AI

Enables semantic vector search using AgentDB for intelligent document retrieval, similarity matching, and context-aware querying in RAG systems.

Enables seamless integration with Replicate AI to run AI models including image generation and large language models.

Generates reusable AI skills with specialist agents optimized via evidence-based prompting for consistent high-quality performance in Claude Code.

Optimizes embedding models and chunking strategies for semantic search and RAG applications in AI systems.

Diagnoses and mitigates context degradation in AI agent systems to maintain performance during large context growth or unexpected failures.

Enables document ingestion, semantic search, and AI context retrieval using Context7 in knowledge management systems.

Advanced system for creating and refining AI skills using deep intent analysis and evidence-based prompt engineering to ensure strategic, well-structured skill design.

Builds a RAG-powered chatbot with integrated UI and backend logic for knowledge retrieval and conversational AI.

Coordinates multiple AI subagents, implements orchestrator patterns, and manages parallel agent workflows for complex task execution.

Scrapes Twitter/X data via GraphQL API with account rotation and session management for data extraction.

Scans invoice directories, extracts key details, and generates categorized statistical reports for transportation, accommodation, and dining invoices.

Generates images using multiple AI models (e.g., ModelScope, Gemini) for seamless integration into other applications.

Provides text-to-speech conversion with script parsing, emotional expression, and post-processing using Microsoft Edge TTS technology.

Specializes in context preservation, compression, and distribution for multi-agent AI workflows and long-running tasks.

Processes CSV files by parsing, transforming, and analyzing data with advanced manipulation features for data-driven tasks.

Builds FiftyOne datasets from media files, imports labels, runs ML inference for vision tasks, and stores predictions.

Embeds interactive Desmos graphs in Obsidian for visualizing mathematical functions, parametric curves, and inequalities with customizable styling.

Provides a token-efficient data encoding format (TOON) for structured data, reducing token usage by 30-60% compared to JSON for LLM applications.

Optimizes AI model responses through advanced prompt design for improved accuracy and application integration.

Creates, edits, and analyzes Excel spreadsheets with formulas, formatting, charts, and data transformation across multiple worksheets.

Optimizes PostgreSQL performance with query tuning, scalable schema design, and performance debugging techniques.

Enables distributed AI system development with AgentDB features including vector search, hybrid search, and QUIC synchronization.

Monitors Claude's responses for difficulty indicators and escalates to specialized agents for tasks exceeding reasoning capacity.

Specialized workflow for developing, training, and deploying machine learning models with expert guidance.