ML Ninjas Platform Agent

The ML Ninjas Platform Agent provides programmatic access to the comprehensive ML education and community platform. Connect to courses, interactive code playground, model zoo, and training infrastructure through the Model Context Protocol.

Overview

This agent enables seamless integration with the ML Ninjas platform, supporting educational workflows, experiment tracking, and model management. Whether you're learning ML fundamentals or deploying production models, this agent provides tools for the entire ML journey.

Key Capabilities

Course Access & Management

• Browse and enroll in ML courses
• Track learning progress across curriculum
• Access course materials and notebooks
• Manage certifications and achievements

Interactive Code Playground

• Execute Python ML code in browser
• Access pre-loaded libraries (NumPy, Pandas, PyTorch, TensorFlow)
• Save and share playground sessions
• Visualize results with Matplotlib and Plotly

Model Zoo Integration

• Browse and search pre-trained models
• Download models across frameworks (PyTorch, TensorFlow, ONNX)
• Access model cards and performance metrics
• Track model versions and updates

Experiment Tracking

• Log training metrics and parameters
• Compare experiments across runs
• Visualize training curves and performance
• Version datasets and model checkpoints

Example Tools

start_training

Start a new training job with experiment tracking.

Code
 
{
  "name": "start_training",
  "description": "Start a model training job with automatic experiment tracking",
  "inputSchema": {
    "type": "object",
    "properties": {
      "experiment_name": {
        "type": "string",
        "description": "Name for this experiment"
      },
      "model_config": {
        "type": "object",
        "description": "Model architecture and hyperparameters"
      },
      "dataset": {
        "type": "string",
        "description": "Dataset identifier or path"
      },
      "hardware": {
        "type": "object",
        "properties": {
          "gpu_count": { "type": "integer" },
          "gpu_type": { "type": "string" },
          "distributed": { "type": "boolean" }
        }
      },
      "tracking": {
        "type": "object",
        "properties": {
          "log_interval": { "type": "integer" },
          "save_checkpoints": { "type": "boolean" },
          "tags": { "type": "array", "items": { "type": "string" } }
        }
      }
    },
    "required": ["experiment_name", "model_config", "dataset"]
  }
}

search_hyperparameters

Run automated hyperparameter search.

Code
 
{
  "name": "search_hyperparameters",
  "description": "Execute hyperparameter search using specified strategy",
  "inputSchema": {
    "type": "object",
    "properties": {
      "base_config": {
        "type": "object",
        "description": "Base model configuration"
      },
      "search_space": {
        "type": "object",
        "description": "Parameter ranges to explore",
        "additionalProperties": {
          "type": "object",
          "properties": {
            "min": { "type": "number" },
            "max": { "type": "number" },
            "distribution": { "type": "string", "enum": ["uniform", "log-uniform", "categorical"] }
          }
        }
      },
      "search_algorithm": {
        "type": "string",
        "enum": ["random", "grid", "bayesian", "hyperband", "optuna"],
        "default": "bayesian"
      },
      "objective_metric": {
        "type": "string",
        "description": "Metric to optimize (e.g., 'val_accuracy')"
      },
      "direction": {
        "type": "string",
        "enum": ["minimize", "maximize"],
        "description": "Optimization direction"
      },
      "max_trials": {
        "type": "integer",
        "default": 100
      },
      "parallel_trials": {
        "type": "integer",
        "default": 4,
        "description": "Number of trials to run in parallel"
      }
    },
    "required": ["base_config", "search_space", "objective_metric"]
  }
}

compare_experiments

Compare metrics across multiple experiments.

Code
 
{
  "name": "compare_experiments",
  "description": "Compare metrics, parameters, and results across experiments",
  "inputSchema": {
    "type": "object",
    "properties": {
      "experiment_ids": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Experiments to compare"
      },
      "metrics": {
        "type": "array",
        "items": { "type": "string" },
        "description": "Metrics to compare (e.g., 'accuracy', 'loss')"
      },
      "visualization_type": {
        "type": "string",
        "enum": ["line_chart", "bar_chart", "parallel_coordinates", "scatter"],
        "default": "line_chart"
      },
      "export_format": {
        "type": "string",
        "enum": ["json", "csv", "pdf"],
        "description": "Format for exported comparison report"
      }
    },
    "required": ["experiment_ids"]
  }
}

resume_training

Resume a paused or failed training job.

Code
 
{
  "name": "resume_training",
  "description": "Resume training from a checkpoint",
  "inputSchema": {
    "type": "object",
    "properties": {
      "experiment_id": {
        "type": "string",
        "description": "Experiment to resume"
      },
      "checkpoint_path": {
        "type": "string",
        "description": "Path to checkpoint file (defaults to latest)"
      },
      "updated_config": {
        "type": "object",
        "description": "Configuration updates (e.g., learning rate)"
      },
      "reset_optimizer": {
        "type": "boolean",
        "default": false,
        "description": "Reset optimizer state"
      }
    },
    "required": ["experiment_id"]
  }
}

Available Resources

• Experiment Logs: All training metrics, parameters, and system logs
• Model Checkpoints: Saved model weights and optimizer states
• Training Datasets: Versioned training and validation datasets
• Hyperparameter Histories: Complete search histories and best configs

Connection Details

Code
 
# MCP Server URL (Placeholder)
mcp://platform.mlninjas.com

# Server Name
mlninjas-platform

# Required Environment Variables
MLNINJAS_API_KEY=your-api-key
MLNINJAS_USER_ID=your-user-id

Example Prompts

Code
 
Show me available courses on Natural Language Processing, my current progress, and
    recommended next steps.

Use Cases

• Learning: "Enroll me in the Deep Learning Fundamentals course"
• Code Execution: "Run this PyTorch model in the playground and show results"
• Model Discovery: "Find pre-trained BERT models for sentiment analysis"
• Training Management: "Track my model training experiments and compare metrics"
• Progress Tracking: "Show my course completion status and certificates"

Next Steps

Related Agents

• BrightPath.ai - ML learning paths and recommendations
• PathX.ai - Algorithm benchmarking and profiling
• BrightForest.ai - MLOps and model deployment