Design Architecture & Service Delivery

At Brainpool, our approach is to deploy models as an API to ensure easy access, speed and scalability.

Design
This is especially impactful in sectors with high quantities of unstructured data with highly bureaucratic policies and processes, such as:
design-architecture
design-architecture
design-architecture
design-architecture

Brainpool Microservices

Through engaging with clients in scoping and consultancy we have realised that the same types of problems arise when dealing with data pipelines, two of which are data streaming and pre-processing. Brainpool has developed microservices that address some of these main issues which can deployed to any type of system infrastructure.
kafka

Data Streaming

Deploy and manage Kafka pipes on Kubernetes and stream from any source to any public-private cloud or/and on-prem servers. Some the key features Data Streamer:

Provisioning of a secure Kafka cluster.

Bespoke broker configuration.

Encrypted communication with an SSL/TSL tunnel.

Monitoring data services via Prometheus dashboard.

Fault tolerant.

At Brainpool, the brightest minds in AI and data science apply proven best practice to maximise the success of AI projects.

pre-processing

Data pre-processing

Deploy and manage a Spark pipeline on Kubernetes. Brainpool have built a Spark pipeline that can pre-process data, for example financial time-series, which can be deployed to any public-private cloud or/and on-prem servers.

Some of the key features:

Scalable pre-processing pipelines.

High-level api’s to popular machine learning technologies, such as, Scala, Java, Python, and R.

Higher-level tools including Spark SQL for SQL and DataFrames, and MLlib for machine learning.

GraphX for graph processing.

Spark Streaming for stream processing.

Fault tolerant.

Some of the open-source tools such as kubernetes and docker make it easy to build a flexible and cloud agnostic pipeline to deploy your AI solution. An example of such pipeline can look like this:
mlops