top of page
  • Writer's pictureMo Sarwat

Releasing Wherobots Spatial Compute & AI Engine 1.0

Updated: Jun 15

Today, we are glad to announce the release of Wherobots Spatial Compute & AI Engine (aka. Wherobots Compute) 1.0 for beta review. Wherobots brings spatial computing and AI technology at the developer' s disposal. Whether you are a data scientist / engineer or ML/AI engineer, you can use Wherobots to develop scalable spatial data analytics applications on your data, processed in your own data stack, and deploy them anywhere. That provides enterprise-scale spatial data infrastructure for myriad applications like automotive, logistics, supply chain, insurance, real estate, agriculture tech, climate tech and more. Below is a summary of features available in 1.0:

Cloud Geospatial Compute & AI
Wherobots 1.0

Same API as Apache Sedona and GeoPandas, plus more

Wherobots Compute shares the same Spatial SQL and Spatial Python API as Apache Sedona to process spatial vector / raster data, and extends it with geospatial ML and AI functions. Wherobots also provides the same GeoPandas API developers use for spatial data processing in Python.Wherobots separates spatial compute from storage, which enables the compute engine to seamlessly integrate with a plethora of data stores. Furthermore, Wherobots is faster, more scalable, and more enterprise-ready compared to Sedona. Wherobots compute also follows the scale as you go model, which is quite suitable for deployment in major public clouds.

No-code deployment in the customer's VPC

Wherobots Compute enables developers to automatically deploy the spatial compute and AI engine to a cluster created in their VPC. In the 1.0 version, users can deploy Wherobots Compute to a Spark Cluster on AWS EMR or Databricks cloud platforms. In future releases, users will be able to deploy Wherobots Compute on various cloud infra. The whole connection and cluster creation / deployment happens via a no-code web interface. For more details, please check out the user manual here

Faster Spatial Data Processing Operations

Wherobots Compute 1.0 can perform spatial data processing operations very efficiently. Compared to OSS Apache Sedona, Wherobots Compute 1.0 executes spatial join operations 10 times faster while utilizing half the compute resources. In addition, users can utilize Wherobots Compute to scale their spatial python data processing pipeline. For instance, running a spatial python data processing pipeline using GeoPandas + Wherobots Compute is at least an order of magnitude more scalable than using a standalone GeoPandas pipeline.

LamPy for scalable spatial machine learning

Wherobots Compute 1.0 comes with a plug and play library, Lampy, for scalable spatial machine learning. LamPy is python library for training and evaluating spatial machine learning models at scale. This will allow users to predict home prices in various neighborhoods, forecast weather, predict foot traffic to buildings, and more... LamPy seamlessly integrates with spatial dataframes processed in Wherobots Compute. That enables users to apply spatial machine learning on their data no matter where it is stored. LamPy supports a plethora of popular spatial statistical analysis and machine learning models such as spatial outlier detection, spatial pattern recognition, spatial hotspot analysis, spatial clustering, spatial regression analysis, spatial autocorrelation, and co-location pattern recognition. Find more information about how to use LamPy here

Wherobots Usage Dashboard

The dashboard allows users to monitor their usage of the Wherobots platform. For instance, it shows statistics on how much spatial data has been read, processed, and written within a Wherobots Compute cluster. The dashboard also provides nuanced statistics on the spatial query and data processing functions invoked for each deployment. Find more information about the dashboard here

164 views0 comments
bottom of page