Engineering Corner

Machine Learning Workflow for Parallel Model Improvements

Deploying machine learning models is hard. Deploying multiple machine-learning models in parallel is harder.

In this MLPOps Community talk, we discuss our strategy for decomposing a machine learning system and providing components that allow highly parallelized development.

Developing a machine learning model

In this post we discuss the methodology and workflow of our ML team and walk through a case study of deploying a real machine learning model at scale.

Avro Serialization with Kafka and AWS Glue

Avro is a popular serialization format for high throughput streaming applications however to get the benefits of improved performance it requires an external schema registry. In this talk, we will explore the challenges of using AWS glue as a schema

Data enrichment and contextual data pipelines

Contextual data is a critical element for a robust ml system yet building real-time data pipelines is very challenging. In this talk, we describe Ironscales' data systems, how we scaled to loads of 50k predictions per second, and how contextual ML models are used in ensemble with other classifiers.