Python SDK

The official Python SDK for the Strongly.AI platform. Pre-installed in all Strongly workspaces.

Overview

The strongly-python SDK provides a simple, unified interface for all Strongly.AI platform features:

• Experiment Tracking - Log parameters, metrics, and artifacts with automatic Git/system info capture
• Model Registry - Publish, deploy, and manage models for production serving
• AutoML - Automated machine learning with AutoGluon integration
• AI Gateway - Access LLMs with completions, chat, streaming, and embeddings

Quick Start

The SDK is pre-installed and auto-configured in Strongly workspaces. No setup required.

import strongly_python as strongly

# Start tracking an experiment
with strongly.start_run(run_name="my-training-run"):
    strongly.log_params({"learning_rate": 0.01, "epochs": 10})

    for epoch in range(10):
        strongly.log_metrics({"loss": 0.5, "accuracy": 0.9}, step=epoch)

    strongly.set_tags(["baseline", "sklearn"])

Features

Experiment Tracking

Track ML experiments with parameters, metrics, artifacts, and automatic system information capture.

import strongly_python as strongly

strongly.set_experiment("my-experiment")

with strongly.start_run(run_name="experiment-1"):
    strongly.log_params({"n_estimators": 100})
    strongly.log_metrics({"accuracy": 0.95})
    strongly.log_artifact("model_report.txt")

Learn more about Experiment Tracking →

Model Registry

Publish, deploy, and manage ML models for production serving.

from sklearn.ensemble import RandomForestClassifier
from strongly_python.mlops import register_model

# Train your model
clf = RandomForestClassifier().fit(X_train, y_train)

# Step 1: Publish to registry
model = register_model(
    name="my-classifier",
    framework="sklearn",
    model=clf
)

# Step 2: Deploy with hardware config
model.deploy(cpu=2, memory_gb=4, wait=True)

Learn more about Model Registry →

AutoML

Train models automatically with AutoGluon integration.

from strongly_python.mlops import automl

job = automl.create_job(
    name="my-model",
    data=df,
    target_column="label",
    problem_type="binary"
)

job.wait()
print(job.get_leaderboard())

Learn more about AutoML →

AI Gateway

Access LLMs through the Strongly AI Gateway.

from strongly_python import gateway

# Text completion
response = gateway.complete("Explain machine learning:")
print(response.content)

# Chat conversation
chat = gateway.Chat()
chat.add_system("You are a helpful assistant.")
response = chat.send("Hello!")

Learn more about AI Gateway →

Autolog Support

Automatically capture training details from popular ML frameworks:

Framework	What's Logged
scikit-learn	Parameters, training score, model
PyTorch	Learning rate, optimizer state
TensorFlow/Keras	Layer config, epoch metrics, model
XGBoost	Parameters, evaluation metrics
LightGBM	Parameters, iteration metrics
CatBoost	Parameters, best scores
Transformers	Training args, final metrics
Optuna	Study info, trial params

import strongly_python as strongly

# Enable autolog for all frameworks
strongly.autolog()

# Or for specific frameworks only
strongly.autolog(include=["sklearn", "xgboost"])