-- Create training_runs table for tracking ML training runs
CREATE TABLE IF NOT EXISTS training_runs (
    id SERIAL PRIMARY KEY,
    run_id VARCHAR(255) NOT NULL UNIQUE,
    model_type VARCHAR(100) NOT NULL,
    experiment_name VARCHAR(255) NOT NULL,
    pipeline_config_hash VARCHAR(64) NOT NULL,
    dataset_version VARCHAR(100),
    metrics_summary JSONB,
    status VARCHAR(50) NOT NULL DEFAULT 'running',
    created_at TIMESTAMP WITH TIME ZONE DEFAULT CURRENT_TIMESTAMP,
    completed_at TIMESTAMP WITH TIME ZONE,
    
    CONSTRAINT valid_status CHECK (status IN ('running', 'completed', 'failed', 'cancelled'))
);

-- Create index on run_id for faster lookups
CREATE INDEX idx_training_runs_run_id ON training_runs(run_id);

-- Create index on experiment_name for filtering by experiment
CREATE INDEX idx_training_runs_experiment ON training_runs(experiment_name);

-- Create index on status for filtering active runs
CREATE INDEX idx_training_runs_status ON training_runs(status);

-- Create index on created_at for chronological queries
CREATE INDEX idx_training_runs_created_at ON training_runs(created_at DESC);