Data Science & ML Agent Skills

Sibyl Theoretical Researcher

Conducts rigorous academic research and generates theoretical proposals with strong mathematical foundations.

Sibyl LaTeX Writer

Converts research papers into NeurIPS LaTeX format and automates the document compilation process.

RAG Engineer

Architects high-performance Retrieval-Augmented Generation systems using advanced embedding, chunking, and search strategies.

Sibyl Supervisor Decision

Analyzes AI experiment results and expert debates to determine whether a project should pivot or proceed.

Sibyl Contrarian

Challenges mainstream assumptions and identifies project blind spots by uncovering counter-evidence through deep academic and web research.

Sibyl Final Critic

Provides rigorous, conference-grade peer reviews for machine learning research papers targeting NeurIPS and ICML standards.

Sibyl Comparative Analyst

Conducts deep comparative research against State-of-the-Art (SOTA) works to define and position project contributions.

Sibyl Pragmatist Research Agent

Generates engineering-focused research proposals by validating academic literature against practical implementation constraints.

Sibyl Experiment Supervisor

Orchestrates machine learning experiments by monitoring GPU availability and dynamically dispatching tasks to remote infrastructure.

Sibyl Literature Researcher

Conducts systematic literature reviews by synthesizing data from arXiv, Google Scholar, bioRxiv, and general web searches.

Sibyl Server Experimenter

Orchestrates and executes machine learning experiments on remote servers using local Claude or Codex agents.

AgentDB Reinforcement Learning

Trains autonomous agents using nine reinforcement learning algorithms to optimize decision-making and agent behavior through experience.

ReasoningBank with AgentDB

Implements high-performance adaptive learning and memory distillation for AI agents using AgentDB's optimized vector database.

Sibyl Empiricist

Facilitates rigorous experimental design and ensures data reproducibility by integrating academic research tools into AI workflows.

D3.js Data Visualization

Creates sophisticated, interactive data visualizations and custom SVG charts using the D3.js library.

Weaviate Vector Database

Streamlines AI-native application development by providing expert patterns for Weaviate vector search, RAG pipelines, and multi-tenancy.

Perplexity Sonar API Integration

Integrates real-time web search and grounded AI responses into applications using the Perplexity Sonar API.

OpenAI SDK Integration

Simplifies the development of AI-powered applications using the OpenAI SDK and the GPT-5 model family.

LangGraph Agent Orchestration

Builds stateful, multi-agent applications and cyclic AI workflows using LangGraph for complex orchestration and persistent state management.

Opus 4.5 Model Migrator

Migrates codebases and prompts from earlier Claude models to Opus 4.5 by automating model string updates and behavioral adjustments.

Instinct Extractor

Analyzes raw interaction logs to extract recurring behavioral patterns and refine AI decision-making.

Market Research & Intelligence

Conducts data-driven market research, competitive analysis, and investor due diligence with verified source attribution.

Advanced Prompt Architect

Generates and optimizes high-performance prompts for Claude, GPT, and other LLMs using research-backed engineering techniques.

Bakken Run Coach

Analyzes fitness training data to provide professional running coaching feedback based on the Bakken Norwegian Model.

Prompt Engineer

Optimizes LLM performance through structured system prompts, few-shot learning, and rigorous evaluation patterns.

Keras Multi-Backend Ops

Standardizes tensor mathematics for BayesFlow extensions using Keras 3's backend-agnostic operations to ensure compatibility across PyTorch, JAX, and TensorFlow.

AI Multimodal Processing

Enables Claude to analyze, transcribe, and generate multimedia content including audio, images, videos, and documents through the Gemini API.

AgentDB Reinforcement Learning

Integrates nine advanced reinforcement learning algorithms into Claude to build self-optimizing autonomous agents that learn from experience.

Multi-Agent System Designer

Architects sophisticated multi-agent systems using proven design patterns, orchestration strategies, and performance evaluation frameworks.

PyHealth AI Toolkit

Develops, tests, and deploys healthcare AI models using clinical data and specialized medical coding systems.