Alternatives to Kubeflow

Why we built an open source, distributed training framework for TensorFlow , Keras , and PyTorch:

At Uber, we apply deep learning across our business; from self-driving research to trip forecasting and fraud prevention, deep learning enables our engineers and data scientists to create better experiences for our users.

TensorFlow has become a preferred deep learning library at Uber for a variety of reasons. To start, the framework is one of the most widely used open source frameworks for deep learning, which makes it easy to onboard new users. It also combines high performance with an ability to tinker with low-level model details—for instance, we can use both high-level APIs, such as Keras, and implement our own custom operators using NVIDIA’s CUDA toolkit.

Uber has introduced Michelangelo (https://eng.uber.com/michelangelo/), an internal ML-as-a-service platform that democratizes machine learning and makes it easy to build and deploy these systems at scale. In this article, we pull back the curtain on Horovod, an open source component of Michelangelo’s deep learning toolkit which makes it easier to start—and speed up—distributed deep learning projects with TensorFlow:

https://eng.uber.com/horovod/

(Direct GitHub repo: https://github.com/uber/horovod)

In mid-2015, Uber began exploring ways to scale ML across the organization, avoiding ML anti-patterns while standardizing workflows and tools. This effort led to Michelangelo.

Michelangelo consists of a mix of open source systems and components built in-house. The primary open sourced components used are HDFS, Spark, Samza, Cassandra, MLLib, XGBoost, and TensorFlow.

!

The algorithms and data infrastructure at Stitch Fix is housed in #AWS. Data acquisition is split between events flowing through Kafka, and periodic snapshots of PostgreSQL DBs. We store data in an Amazon S3 based data warehouse. Apache Spark on Yarn is our tool of choice for data movement and #ETL. Because our storage layer (s3) is decoupled from our processing layer, we are able to scale our compute environment very elastically. We have several semi-permanent, autoscaling Yarn clusters running to serve our data processing needs. While the bulk of our compute infrastructure is dedicated to algorithmic processing, we also implemented Presto for adhoc queries and dashboards.

Beyond data movement and ETL, most #ML centric jobs (e.g. model training and execution) run in a similarly elastic environment as containers running Python and R code on Amazon EC2 Container Service clusters. The execution of batch jobs on top of ECS is managed by Flotilla, a service we built in house and open sourced (see https://github.com/stitchfix/flotilla-os).

At Stitch Fix, algorithmic integrations are pervasive across the business. We have dozens of data products actively integrated systems. That requires serving layer that is robust, agile, flexible, and allows for self-service. Models produced on Flotilla are packaged for deployment in production using Khan, another framework we've developed internally. Khan provides our data scientists the ability to quickly productionize those models they've developed with open source frameworks in Python 3 (e.g. PyTorch, sklearn), by automatically packaging them as Docker containers and deploying to Amazon ECS. This provides our data scientist a one-click method of getting from their algorithms to production. We then integrate those deployments into a service mesh, which allows us to A/B test various implementations in our product.

For more info:

• Our Algorithms Tour: https://algorithms-tour.stitchfix.com/
• Our blog: https://multithreaded.stitchfix.com/blog/
• Careers: https://multithreaded.stitchfix.com/careers/

#DataScience #DataStack #Data

Why we built Marmaray, an open source generic data ingestion and dispersal framework and library for Apache Hadoop :

Built and designed by our Hadoop Platform team, Marmaray is a plug-in-based framework built on top of the Hadoop ecosystem. Users can add support to ingest data from any source and disperse to any sink leveraging the use of Apache Spark . The name, Marmaray, comes from a tunnel in Turkey connecting Europe and Asia. Similarly, we envisioned Marmaray within Uber as a pipeline connecting data from any source to any sink depending on customer preference:

https://eng.uber.com/marmaray-hadoop-ingestion-open-source/

(Direct GitHub repo: https://github.com/uber/marmaray Kafka Kafka Manager )

I already use DVC to keep track and store my datasets in my machine learning pipeline. I have also started to use MLflow to keep track of my experiments. However, I still don't know whether to use DVC for my model files or I use the MLflow artifact store for this purpose. Or maybe these two serve different purposes, and it may be good to do both! Can anyone help, please?

Can you please advise which one to choose FastText Or Gensim, in terms of:

1. Operability with ML Ops tools such as MLflow, Kubeflow, etc.
2. Performance
3. Customization of Intermediate steps
4. FastText and Gensim both have the same underlying libraries
5. Use cases each one tries to solve
6. Unsupervised Vs Supervised dimensions
7. Ease of Use.

Please mention any other points that I may have missed here.

I am working on a project that grabs a set of input data from AWS S3, pre-processes and divvies it up, spins up 10K batch containers to process the divvied data in parallel on AWS Batch, post-aggregates the data, and pushes it to S3.

I already have software patterns from other projects for Airflow + Batch but have not dealt with the scaling factors of 10k parallel tasks. Airflow is nice since I can look at which tasks failed and retry a task after debugging. But dealing with that many tasks on one Airflow EC2 instance seems like a barrier. Another option would be to have one task that kicks off the 10k containers and monitors it from there.

I have no experience with AWS Step Functions but have heard it's AWS's Airflow. There looks to be plenty of patterns online for Step Functions + Batch. Do Step Functions seem like a good path to check out for my use case? Do you get the same insights on failing jobs / ability to retry tasks as you do with Airflow?

I am looking for an open-source scheduler tool with cross-functional application dependencies. Some of the tasks I am looking to schedule are as follows:

1. Trigger Matillion ETL loads
2. Trigger Attunity Replication tasks that have downstream ETL loads
3. Trigger Golden gate Replication Tasks
4. Shell scripts, wrappers, file watchers
5. Event-driven schedules

I have used Airflow in the past, and I know we need to create DAGs for each pipeline. I am not familiar with Jenkins, but I know it works with configuration without much underlying code. I want to evaluate both and appreciate any advise

How Uber developed the open source, end-to-end distributed tracing Jaeger , now a CNCF project:

Distributed tracing is quickly becoming a must-have component in the tools that organizations use to monitor their complex, microservice-based architectures. At Uber, our open source distributed tracing system Jaeger saw large-scale internal adoption throughout 2016, integrated into hundreds of microservices and now recording thousands of traces every second.

Here is the story of how we got here, from investigating off-the-shelf solutions like Zipkin, to why we switched from pull to push architecture, and how distributed tracing will continue to evolve:

https://eng.uber.com/distributed-tracing/

(GitHub Pages : https://www.jaegertracing.io/, GitHub: https://github.com/jaegertracing/jaeger)

Bindings/Operator: Python Java Node.js Go C++ Kubernetes JavaScript OpenShift C# Apache Spark

Our first experience with .NET core was when we developed our OSS feature management platform - Tweek (https://github.com/soluto/tweek). We wanted to create a solution that is able to run anywhere (super important for OSS), has excellent performance characteristics and can fit in a multi-container architecture. We decided to implement our rule engine processor in F# , our main service was implemented in C# and other components were built using JavaScript / TypeScript and Go.

Visual Studio Code worked really well for us as well, it worked well with all our polyglot services and the .Net core integration had great cross-platform developer experience (to be fair, F# was a bit trickier) - actually, each of our team members used a different OS (Ubuntu, macos, windows). Our production deployment ran for a time on Docker Swarm until we've decided to adopt Kubernetes with almost seamless migration process.

After our positive experience of running .Net core workloads in containers and developing Tweek's .Net services on non-windows machines, C# had gained back some of its popularity (originally lost to Node.js), and other teams have been using it for developing microservices, k8s sidecars (like https://github.com/Soluto/airbag), cli tools, serverless functions and other projects...