Heimdall offers a free tier with optional paid features.

Heimdall

A cloud workspace that carries prepared data into model training, forecasting, and production prediction APIs.

freemium

Last Updated 8/28/2026

About Heimdall

Heimdall is a browser-based cloud data and machine-learning platform that connects data preparation with model deployment. Teams can bring CSV files, database tables, warehouse snapshots, and unstructured archives into Lake, organize them through bronze, silver, and gold stages, then work with the same catalog in Lab or the no-code modeling products. Lab provides Python notebooks with pandas and common charting libraries, while ML handles classification and regression, Forecast builds time-series models, and Loop supports models that learn from feedback. Trained models can be exposed as REST APIs and monitored in the same workspace. For a small team, that continuity can remove much of the routine data movement between storage, notebooks, training systems, and model servers. It is less suitable for buyers who need clearly published capacity limits, broad integration guarantees, or detailed enterprise controls before evaluation, because those details are not stated in the available product material.

Best for

Small data teams that want storage, notebooks, no-code modeling, and API deployment in one browser workspace.

Analysts who can prepare datasets but prefer guided model training over maintaining a custom ML stack.

Developers who need a direct route from a curated dataset to a callable REST prediction endpoint.

Best use cases

A team consolidates CSV exports and database snapshots in Lake, curates a gold table, and trains a production model.

An analyst starts with a dated numeric series, compares forecast metrics and confidence intervals, and deploys future values.

A Python user explores bronze or silver tables in Lab, charts results, and saves a non-destructive silver transformation.

Compare before choosing

Compared with separate lake, notebook, training, and serving tools, Heimdall keeps the workflow in one cataloged platform.

Compared with code-first ML stacks, its guided ML and Forecast paths reduce setup work but expose less implementation control.

Compared with products that publish detailed tiers, Heimdall gives buyers less cost and capacity information before signup.

Features

Medallion data workflow

Catalog raw inputs in Lake, refine bronze and silver data, and publish gold datasets for modeling.

No-code forecasting

Train forecasts from time-series data, inspect error metrics and confidence intervals, and deploy an API.

Browser Python notebooks

Analyze Lake tables with pandas and plotting libraries, then save results as new silver datasets.

Pros

One catalog connects data ingestion, preparation, model training, deployment, and monitoring in a continuous workflow.
No-code ML and Forecast paths can shorten the route from a prepared gold dataset to a working prediction endpoint.
Lab supports pandas and familiar plotting libraries while saving transformations without overwriting the source table.

Cons

Exact paid prices, free-tier quotas, and billing periods are not published, which makes budget planning difficult.
The gold-dataset workflow adds preparation steps before teams can train models on their own production data.
Public materials provide limited detail on integration breadth and enterprise controls, so fit may be hard to judge early.