Data Science & ML Agent Skills

ReasoningBank Intelligence

Equips AI agents with adaptive learning and pattern recognition capabilities to optimize strategies based on historical performance.

ReasoningBank Adaptive Learning

Implements high-performance adaptive learning and memory distillation for AI agents using AgentDB’s ultra-fast vector database.

AgentDB Memory Patterns

Implements high-performance persistent memory and learning patterns for stateful AI agents using AgentDB.

Flow Nexus Neural Networks

Trains, deploys, and manages distributed neural networks across isolated E2B sandbox environments using advanced machine learning architectures.

Seedance AI Video Generation

Generates cinematic AI videos with synchronized audio and precision lip-syncing using the Seedance 1.5 Pro model.

Nanobanana Image Generation

Generates and edits high-quality images using Google’s Nanobanana model via the Replicate API.

LA-Bench Procedure Generator

Generates detailed experimental procedures from LA-Bench datasets by orchestrating a multi-agent workflow for parsing, reference fetching, and quality validation.

D3.js Visualization

Generates sophisticated, interactive data visualizations and custom SVG graphics with precise control over visual elements and transitions.

Sprint 1 Executor

Executes structured, atomic tasks for fantasy football analytics, FASA optimization, and trade intelligence within a standardized sprint framework.

Notebook Creator

Generates standardized Jupyter notebooks for fantasy football data analysis, integrating DuckDB connections, dbt mart queries, and professional visualization patterns.

CellPhoneDB Analysis

Facilitates single-cell RNA-seq analysis of cell-cell communication and ligand-receptor interactions using the CellPhoneDB framework.

scArches

Streamlines single-cell omics analysis through deep learning-based data integration, reference mapping, and model surgery techniques.

Vector RAG Tool

Indexes codebases and documents for semantic search to enable local Retrieval-Augmented Generation (RAG) using Ollama and FAISS.

Gemini Google Maps Tool

Integrates Gemini AI with Google Maps grounding to provide accurate, location-aware responses and recommendations for over 250 million places worldwide.

CrewAI

Facilitates the development and orchestration of collaborative multi-agent AI systems and automated workflows.

Basilica CLI Helper

Rent high-performance GPUs and manage machine learning training jobs on Basilica's decentralized compute marketplace.

LlamaFarm Config Validator

Validates LlamaFarm YAML configurations against schemas and custom business rules to ensure project integrity.

Example Analysis Skill

Provides a standardized workflow for data preparation, exploratory analysis, and high-quality visualization using Python libraries.

Vector RAG GUI

Search and synthesize local vector RAG stores using an AI-powered graphical interface.

Complex Tensor Handler

Manages complex-valued PyTorch tensors for astronomical imaging, signal processing, and frequency-domain neural networks.

PyTorch Shape Validator

Validates and documents PyTorch tensor shapes through inline comments and runtime assertions to prevent common deep learning bugs.

Architecture Reviewer

Enforces adherence to PRISM architectural patterns by validating the implementation of Runner and Trainer classes during refactoring.

Memory Leak Detector

Identifies and resolves memory exhaustion issues in Python and PyTorch applications, specifically targeting matplotlib figures and tensor accumulation.

LLM Integration

Configures and manages diverse LLM providers, streaming callbacks, and token optimization strategies for XSky-based agentic workflows.

LLM CLI

Interacts with diverse large language models through a command-line interface to perform tasks like prompt execution, data extraction, and embedding management.

Senior Prompt Engineer

Optimizes LLM performance through advanced prompt engineering, RAG architecture design, and agentic system orchestration.

Cleaning Data

Implements a systematic data quality remediation process to detect duplicates, handle outliers, and standardize inconsistencies for reliable analysis.

Creating Visualizations

Generates professional terminal-based and image-based visualizations to communicate data patterns and analytics results clearly.

Presenting Data

Generates professional, data-driven presentations and technical whitepapers with robust citation management and reproducibility standards.

Writing Plans for LLM Training

Generates structured, micro-task-based implementation plans specifically for large language model fine-tuning and NLP workflows.