Tool and Framework Expertise Questions

import mlflow
import mlflow.sklearn
from sklearn.datasets import load_iris
from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score
import joblib

X, y = load_iris(return_X_y=True)
X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42)

with mlflow.start_run():
    # log a hyperparameter
    mlflow.log_param('C', 1.0)

    model = LogisticRegression(C=1.0, max_iter=100)
    # pretend training loop (scikit-learn does whole-fit at once)
    model.fit(X_train, y_train)

    # log one metric (validation accuracy)
    preds = model.predict(X_val)
    acc = accuracy_score(y_val, preds)
    mlflow.log_metric('val_accuracy', float(acc))

    # save and log model artifact
    joblib.dump(model, 'model.joblib')
    mlflow.log_artifact('model.joblib')

import tensorflow as tf
from tensorflow import keras
import numpy as np

# Example model
model = keras.Sequential([
    keras.layers.Input(shape=(4,)),
    keras.layers.Dense(8, activation='relu'),
    keras.layers.Dense(3, activation='softmax')
])
model.compile(optimizer='adam', loss='sparse_categorical_crossentropy')

# Dummy training (optional)
x = np.random.rand(100, 4)
y = np.random.randint(0, 3, size=(100,))
model.fit(x, y, epochs=1)

# Export as SavedModel with default serving signature
export_path = "exported_models/iris/1"  # versioned directory for TF Serving
tf.saved_model.save(model, export_path, signatures=model.call.get_concrete_function(
    tf.TensorSpec(shape=[None, 4], dtype=tf.float32, name="input")
))
# Alternatively: model.save(export_path, include_optimizer=False, save_format="tf")

# JSON payload: instances is a list of input rows
curl -X POST http://localhost:8501/v1/models/iris:predict \
  -H "Content-Type: application/json" \
  -d '{"instances": [[5.1, 3.5, 1.4, 0.2], [6.2, 3.4, 5.4, 2.3]]}'