Feast: The open source feature store for AI

https://www.redhat.com/en/blog/feast-open-source-feature-store-ai

#RedHat #OpenSource #Feast #FeatureStore #AI #DataScience

💬 Discussion fuel:

How’s your team handling feature sprawl?
Tried Iceberg/Nessie? Hot takes?

#MLOps #DataEngineering #FeatureStore

Part2: #dailyreport #devops #featurestore #dataplatform
enable quicker innovation
3) CI - to solve raiseed operational challenges from 1)
and 2). Developers regularly merge their code changes
into a central repository, after which automated builds
and tests are run.
4) Infrastructure automation - infrastructure as code and
configuration management, help to keep computing
resources elastic and responsive to frequent changes.
5) monitoring and logging - helps engineers track the
performance of applications and infrastructure so they
can react quickly to problems.
6) Communication and Collaboration - by development and
operations, around information sharing and facilitating
communication through the use of chat applications,
issue or project tracking systems, and wikis.

DevOps lifecycle https://www.tecton.ai/wp-content/uploads/2020/04/techblog-img1@2x-1024x439.png
Best Article https://www.tecton.ai/blog/devops-ml-data/
蠡

Part1: #dailyreport #devops #featurestore #dataplatform
I have been reading Uber blog about Data Platform with
Feature Store and DevOps for ML.

Modern Data platform architecture:
sources -> Staging_area and data lake -> Warehouse ->
(Feature Store -> models), (data martes -> users)

For streaming: sources -> Kafka, Kinesis -> models, users,
(mart -> users)

DevOps is about removing the barriers between two
traditionally siloed teams, development and
operations. And reduce time to market.

DevOps practices:
1) Continuous Delivery - perform very frequent but small
updates - faster to market, less risky development
2) microservices architecture - for more flexible and