3. Data, Science & AI

Orchestrates AI agent and skill structures to maintain documented, consistent project extensions in LLM-based systems.

Monitors and diagnoses AI model health to ensure optimal performance and reliability in production environments.

Coordinates multiple AI agents to execute complex workflows through dynamic task delegation and synchronization.

Expertise in integrating Chroma vector database for semantic search, embeddings, and RAG systems in AI applications.

Manages end-to-end data processing pipelines for efficient data transformation and integration.

Advanced coordination of multiple AI agents with built-in conflict resolution for seamless collaborative task execution.

Queries KBase/BERDL biological data lake via REST API for table exploration, schema retrieval, and SQL-based data analysis.

Optimizes prompts for Claude Opus 4.5, covering system tuning, version migration, and agentic workflow debugging.

Creates data analysis reports with all quantitative information displayed via programmatically generated visualizations, preventing fabricated numbers and ensuring data accuracy.

Enables development of autonomous AI agents in Python via Claude Agent SDK with GTVR execution, multi-agent coordination, and production deployment strategies.

Enables creation of conversational AI agents and multi-agent systems using OpenAI Agents SDK with function tools for tool-calling patterns.

Guides implementation of data pipelines with SQLAlchemy 2.0 patterns and warehouse architectures (Kimball, medallion layers) using Python and PostgreSQL.

Provides hybrid document search using exact text matching and semantic embeddings for comprehensive document discovery.

Enables continuous system improvement by extracting patterns from user sessions and dynamically updating capabilities.

Determines when to ask, proceed, or defer in AI workflows based on confidence levels and decision reversibility.

Adds new OpenCode skills or subagents with strict permissions and structured composition for AI agent systems.

Analyzes UI screenshots with Gemini Vision AI to detect elements, generate bounding boxes, and produce JSON/annotated images for UI analysis.

Enables building conversational AI agents with external tool access using OpenAI Agents SDK and Model Context Protocol (MCP) for multi-provider backend integration.

Converts natural language plans into Ralph's structured JSON format for autonomous agent workflows. Streamlines agent setup by standardizing input.

Configures PostgreSQL databases for FastAPI applications using Docker and environment variables, including setup scripts and schema handling.

Builds flexible database queries with multi-condition filtering, sorting, and pagination for API endpoints using SQLModel.

Generates new Claude Code agents (subagents) with best practices for specialized AI assistants in project development.

Verifies the /notebooklm-generate command functionality for creating AI/ML notebooks using a sample URL.

Enables LLM fine-tuning with modern techniques (QLoRA, full fine-tuning) using Hugging Face TRL and PEFT, covering dataset prep to evaluation.