STAMM is an open-source framework for deploying and operating machine-learning soft sensors in industrial processes. It monitors live process data for concept drift and changes in operating regimes, and its dashboard brings together process measurements, sensor outputs, drift signals, and historical context. People can add labels through the dashboard for human review. A model registry records versions, configurations, artifacts, validation results, and metadata. STAMM serves Python and R soft sensors through REST inference; its event-driven orchestrator calls the registry over REST and links predictions to the data snapshot behind each result. The drift detector package includes 10 detectors behind a shared interface and can also be used separately. Data can enter through equipment REST hooks, MQTT applications, or frameworks such as LEAF. The reference time-series store is InfluxDB, while a PostgreSQL adapter is in progress. The project is licensed under Apache License 2.0 and offers self-hosted deployment. Its reference demo uses an industrial-scale penicillin fermentation simulator, with a curated dataset, Node-RED bioreactor, and working model registry. The project surfaces when maintenance may be needed, but does not prescribe how to rebuild a model.
Who it is for
STAMM is suited to process modelers and ML engineers deploying soft sensors alongside physical instruments in live industrial systems. Operators and project or production managers can use its dashboard to review process measurements, model outputs, and drift signals.
What is good
- Open-source under Apache License 2.0.
- Monitors live data for drift and operating-regime changes.
- Registry tracks versions, artifacts, validation results, and metadata.
- Dashboard supports historical context and human-in-the-loop labelling.
- Ten drift detectors share a common interface.
- Self-hosted deployment is available.
What to know first
- A PostgreSQL adapter is still in progress.
- The framework flags possible maintenance needs but does not guide model rebuilding.
- Docker 24 or later and Docker Compose are required for the reference deployment.
- At least 8 GB RAM is recommended.
Verdict
Choose STAMM if you need an open-source, self-hosted framework to monitor industrial soft sensors and track model versions and results. Its drift signals can identify when maintenance may be needed, but the framework does not prescribe how to rebuild a model.
Get started with STAMM
- Visit the STAMM project website.
- Prepare Linux, macOS, or Windows with Docker 24 or later and Docker Compose.
- Allow at least 8 GB of RAM for the recommended setup.
- Use the reference Docker Compose deployment.
- Explore the demo, which includes a fermentation simulator, curated dataset, Node-RED bioreactor, and model registry.
Limits to know first
The reference deployment recommends 8 GB RAM or more and requires Docker 24 or later with Docker Compose. STAMM surfaces possible maintenance needs but does not prescribe model rebuilding; its PostgreSQL adapter is in progress.
Questions about STAMM
Is STAMM free?
Yes. It is listed as free and open source, and is released under Apache License 2.0.
Where can STAMM run?
Its listed platforms are API, Linux, macOS, self-hosted, web, and Windows. The reference Docker Compose installation supports Linux, macOS, or Windows and requires Docker 24 or later.
Which soft-sensor languages does it support?
The maker describes support for Python and R soft sensors, served through REST inference.
What does its model registry track?
It tracks model versions, configuration, artifacts, validation results, and metadata.
What data sources and storage does it describe?
Data can arrive through equipment REST hooks, MQTT applications, or frameworks such as LEAF. The reference time-series store is InfluxDB; a PostgreSQL adapter is in progress.
Who is the project contact?
The project page lists David Camilo Corrales at INRAE, Toulouse Biotechnology Institute, as a contact.
Compared on machine learning model monitoring software
- Drift monitoring
- Yes
- Model performance metrics
- Yes
- Deployment options
- self-hosted
