blurr-dev 0.1073

Data aggregation pipeline for running real-time predictive models

Table of contents

• What is Blurr
• Is Blurr for you?
• Blurr is MLOps
• Tutorial & Docs
• Try Blurr
• Contribute
• Data Science 'Joel Test'
• Roadmap

What is Blurr?

Blurr transforms structured, streaming raw data into features for model training and prediction using a high-level expressive YAML-based language called the Blurr Transform Spec (BTS).

The BTS is a data transform definition for structured data. The BTS encapsulates the business logic of data transforms and Blurr orchestrates the execution of data transforms. Blurr unner-agnostic, so BTSs can be run by event processors such as Spark, Spark Streaming or Flink.

This looks like any other ETL pipeline. At this point, Blurr doesn't do anything special that you cannot do with Spark, for instance. Blurr shines when an offline model pipeline needs to be turned into an online scoring pipeline.

Is Blurr for you?

Blurr is for you if:

1. You are well on your way on the ML 'curve of enlightenment', and are thinking about how to do online scoring

2. You self-identify as a data scientist, a data engineer, or an ML engineer. But you believe that these distinctions are temporary. With the right tools, these are all one person. data science, operations, and engineering working together with minimal dependencies is critical to success of production ML efforts.

Blurr is MLOps

Blurr is a collection of components built for MLOps, the Blurr Core library is one of them. Blurr Core ⊆ Blurr

We believe in a world where everyone is a data engineer. Or a data scientist. Or an ML engineer. The lines are blurred (cough). Just like development and operations became DevOps over time

We see a future where MLOps means teams putting together various technologies to suit their needs. For production ML applications, the speed of experimentation and iterations is the difference between success and failure. The BTS helps teams iterate on features faster. The vision for Blurr is to build MLOps components to help ML teams experiment at high speed.

How to build AI culture: go through the curve of enlightenment

Tutorial and Docs

Coming up with features is difficult, time-consuming, requires expert knowledge. 'Applied machine learning' is basically feature engineering --- Andrew Ng

Read the docs

Streaming BTS Tutorial | Window BTS Tutorial

Preparing data for specific use cases using Blurr:

• Dynamic in-game offers (Offer AI)
• Frequently Bought Together
• New York Stock Exchange Prediction

Try Blurr

One way to interact with Blurr is by using a Command Line Interface (CLI). The CLI is used to run blurr locally and is a great way of validating and testing the BTSs before deploying them in production.

$ pip install blurr

Transform data

$ blurr transform \
     --streaming-dtc ./dtcs/sessionize-dtc.yml \
     --window-dtc ./dtcs/windowing-dtc.yml \
     --source file://path

CLI documentation

Contribute to Blurr

Welcome to the Blurr community! We are so glad that you share our passion for building MLOps!

Please create a new issue to begin a discussion. Alternatively, feel free to pick up an existing issue!

Please sign the Contributor License Agreement before raising a pull request.

Data Science 'Joel Test'

Inspired by the (old school) Joel Test to rate software teams, here's our version for data science teams. What's your score?

1. Data pipelines are versioned and reproducible
2. Pipelines (re)build in one step
3. Deploying to production needs minimal engineering help
4. Successful ML is a long game. You play it like it is
5. Kaizen. Experimentation and iterations are a way of life

Roadmap

Blurr is currently in Developer Preview. Stay in touch!: Star this project or email [e-mail]

• Local transformations only
• Support for custom functions and other python libraries in the BTS
• Spark runner
• S3 support for data sink
• DynamoDB as an Intermediate Store
• Features server