Pulsar
What it is
Pulsar is a free, open-source community edition for monitoring machine-learning models in production. It tracks model drift, concept drift, and model performance, using training data as a baseline. An SDK and containerised ingestion service collect model inputs and outputs into a time-series database, supporting historical and real-time data. The current database integration is InfluxDB, and the system is designed for large ingestion volumes without degraded performance. A scheduler can trigger standard or customised alerts when metrics reach thresholds. Grafana displays drift measures across time ranges and user-defined features. Listed drift measures include Kullback-Leibler divergence, Kolmogorov-Smirnov, Levene, and Wasserstein tests; performance measures include accuracy, recall, F1, Brier score, mean squared error, and R2. Pulsar is self-hosted, with deployment logos including Flask, FastAPI, TensorFlow Extended, Seldon, AWS, Azure, Google Cloud, and Grafana.
For and against
For
- Monitors drift and model performance
- Supports historical and real-time data
- Includes Grafana visualization
- Offers standard and customised alerts
- Provides many statistical and performance metrics
Against
- Current time-series integration is InfluxDB
- Deployment is self-hosted
- Cloud marketplace augmentation is still being developed
The verdict
Pulsar brings model monitoring, metric calculation, alerting, and Grafana views into a self-hosted community edition. Its current database integration and deployment model should guide setup decisions.
Who it suits
Pulsar is aimed at software engineers, data scientists, and machine-learning engineers monitoring deployed models. It suits teams seeking self-hosted ingestion, dashboards, metrics, and threshold alerts.
On the record
- Additional integrations
- The homepage displays Flask, FastAPI, TensorFlow, Seldon, AWS, Azure, Google Cloud, and Grafana as deployment integration logos.pulsar.ml · read 4 Oct 2026
- Alerts
- Pulsar describes scheduled monitoring that can trigger standard and customized alerts when metrics reach thresholds.pulsar.ml · read 7 Oct 2026
- Audience
- The maker identifies software engineers, data scientists, and machine learning engineers as users of Pulsar.pulsar.ml · read 7 Oct 2026
- Capacity
- The features page says Pulsar is built to handle large volumes of ingested data without degraded performance.pulsar.ml · read 4 Oct 2026
- Cloud marketplaces
- The FAQ says Pulsar is being developed to augment Azure and AWS marketplaces.pulsar.ml · read 4 Oct 2026
- Community edition
- The site describes Pulsar as a 100% community edition.pulsar.ml · read 7 Oct 2026
- Dashboard
- Pulsar includes Grafana for visualising data drift metrics over different time ranges for user-defined features.pulsar.ml · read 4 Oct 2026
- Database integration
- The site says its current time series database integration is InfluxDB, with support for other time series databases planned.pulsar.ml · read 7 Oct 2026
- Data ingestion
- Pulsar ingests model inputs and outputs into a time series database and supports historical and real-time data.pulsar.ml · read 7 Oct 2026
- Deployment
- Pulsar provides an SDK and a containerized data ingestion service for integration with a user's machine and production code.pulsar.ml · read 7 Oct 2026
- Deployment integrations
- The homepage displays Flask, FastAPI, TensorFlow Extended, Seldon, AWS, Azure, Google Cloud, and Grafana as deployment integration logos.pulsar.ml · read 7 Oct 2026
- Drift metrics
- Listed data drift metrics include Kullback-Leibler divergence, Kolmogorov-Smirnov, Levene, Wasserstein, and other statistical tests.pulsar.ml · read 7 Oct 2026
- Expert services
- Pulsar Experts are described as machine learning engineers who can help users learn Pulsar or build machine learning pipelines.pulsar.ml · read 7 Oct 2026
- Maker
- Rocket Science Development is identified as Pulsar.ml's maker, and its site lists an address in Longueuil, Québec, Canada.rocketscience.one · read 7 Oct 2026
- Metrics
- Pulsar calculates data drift and model performance metrics using training data as a baseline.pulsar.ml · read 4 Oct 2026
- Metrics dashboard
- Pulsar includes Grafana for visualizing data drift metrics over time ranges and user-defined features.pulsar.ml · read 7 Oct 2026
- Performance metrics
- Listed performance metrics include accuracy, recall, F1 score, Brier score, mean squared error, and R2 score.pulsar.ml · read 7 Oct 2026
- Purpose
- Pulsar monitors machine learning models in production, including data and model drift and model performance.pulsar.ml · read 7 Oct 2026
