Jan 26, 2018 · newsletter

Serverless data science

Cloud computing has transformed enterprise IT. It makes it possible to rapidly scale up and down a complex and global infrastructure, without the overhead of a datacenter. But living in the cloud can be expensive, and you still need to maintain computers and their operating systems, even if they are virtual. That’s why we’ve been watching with interest the rise of “serverless” computing.

Serverless has the potential to open up “big data” work to mere mortal data scientists who don’t have the budget or engineering support for a long-lived analytics cluster, and to make life simpler and reduce costs for those that do.

Traditional cloud “elastic compute” systems (like Amazon’s EC2, Google’s Computer Engine, or Azure’s Virtual Machines) allow you to run applications without maintaining hardware. The goal of “serverless” is to go even further, and make it possible to run applications without worrying about hardware or the operating system.

Serverless environments (like Amazon’s Lambda, Google’s Cloud Functions, or Azure’s Functions) can be thought of as computing environments that pop into existence for the duration of a single function call, and are then destroyed. Configuring and maintaining the underlying operating system is somebody else’s problem.

Because the serverless instance exists only for the duration of the function, there’s no idle time and your bill scales almost perfectly with the amount of compute you use. Combined with the fact that you no longer need to configure and maintain the operating system, this can result in big savings. For example, our friends at Postlight converted their Readability application to run on AWS Lambda and reduced the monthly cost from over $10,000 to $370.

But it’s not all good news. Because the environment ends after the function finishes, input and output must occur via a web API or a database connection. There is no internal state or disk. And the various cloud providers place CPU, RAM, time, and programming language constraints on what you can do. (For example, AWS Lambda functions must run Python, C#, node.js or Java; R is not an option. And the function must return in less than 300 seconds and use no more than 1.5Gb of RAM.)

These limitations might seem to make serverless less appealing to power-hungry data scientists and data engineers - and indeed, so far most of the prominent uses of serverless have been in web applications, where the computational requirements are less intense. But in many ways the constraints of serverless are similar to those faced in distributed analytics clusters running Hadoop or Spark. To do data analytics in these environments, we have the map-reduce computing paradigm, which parallelizes work by splitting it into small parcels.

PyWren’s computational power can grow to many TFLOPS as the number of workers (inexpensive AWS Lambda instances in this case) increases. Image credit: Occupy the Cloud: Distributed Computing for the 99%

PyWren is a distributed analytics tool that connects the dots. It splits up large analytics jobs into smaller parcels of work, and ships them off to hundreds or even thousands of serverless AWS Lambda instances. This makes PyWren (with AWS Lambda as a computational “backend”) a light-weight alternative to complex, expensive (and admittedly more robust) map-reduce frameworks such as Hadoop and Spark.

And - perhaps most intriguing for us at Cloudera Fast Forward Labs, given our interest in machine learning - it’s exciting to see serverless used to speed up hyperparamter optimization, a relatively simple (but computationally intensive) part of model building.

For more on PyWren, see Occupy the Cloud: Distributed Computing for the 99%. For more on the implications of serverless more generally, see Serverless Computing: Economic and Architectural Impact.

Read more

Feb 14, 2018 · fast forward food labs
Jan 26, 2018 · newsletter

Latest posts

Nov 15, 2020 · post

Representation Learning 101 for Software Engineers

by Victor Dibia · Figure 1: Overview of representation learning methods. TLDR; Good representations of data (e.g., text, images) are critical for solving many tasks (e.g., search or recommendations). Deep representation learning yields state of the art results when used to create these representations. In this article, we review methods for representation learning and walk through an example using pretrained models. Introduction Deep Neural Networks (DNNs) have become a particularly useful tool in building intelligent systems that simplify cognitive tasks for users. more
Jun 22, 2020 · post

How to Explain HuggingFace BERT for Question Answering NLP Models with TF 2.0

by Victor · Given a question and a passage, the task of Question Answering (QA) focuses on identifying the exact span within the passage that answers the question. Figure 1: In this sample, a BERTbase model gets the answer correct (Achaemenid Persia). Model gradients show that the token “subordinate ..” is impactful in the selection of an answer to the question “Macedonia was under the rule of which country?". This makes sense .. good for BERTbase. more
Jun 16, 2020 · notebook

Evaluating QA: Metrics, Predictions, and the Null Response →

by Melanie · A deep dive into computing QA predictions and when to tell BERT to zip it! In our last post, Building a QA System with BERT on Wikipedia, we used the HuggingFace framework to train BERT on the SQuAD2.0 dataset and built a simple QA system on top of the Wikipedia search engine. This time, we’ll look at how to assess the quality of a BERT-like model for Question Answering.
May 19, 2020 · notebook

Building a QA System with BERT on Wikipedia →

by Melanie · So you’ve decided to build a QA system. You want to start with something simple and general so you plan to make it open domain using Wikipedia as a corpus for answering questions. You want to use the best NLP that your compute resources allow (you’re lucky enough to have access to a GPU) so you’re going to focus on the big, flashy Transformer models that are all the rage these days.
Apr 28, 2020 · notebook

Intro to Automated Question Answering →

by Melanie · Welcome to the first edition of the Cloudera Fast Forward blog on Natural Language Processing for Question Answering! Throughout this series, we’ll build a Question Answering (QA) system with off-the-shelf algorithms and libraries and blog about our process and what we find along the way. We hope to wind up with a beginning-to-end documentary that provides:
Apr 1, 2020 · newsletter

Enterprise Grade ML

by Shioulin · At Cloudera Fast Forward, one of the mechanisms we use to tightly couple machine learning research with application is through application development projects for both internal and external clients. The problems we tackle in these projects are wide ranging and cut across various industries; the end goal is a production system that translates data into business impact. What is Enterprise Grade Machine Learning? Enterprise grade ML, a term mentioned in a paper put forth by Microsoft, refers to ML applications where there is a high level of scrutiny for data handling, model fairness, user privacy, and debuggability. more

Popular posts

Oct 30, 2019 · newsletter
Exciting Applications of Graph Neural Networks
Nov 14, 2018 · post
Federated learning: distributed machine learning with data locality and privacy
Apr 10, 2018 · post
PyTorch for Recommenders 101
Oct 4, 2017 · post
First Look: Using Three.js for 2D Data Visualization
Aug 22, 2016 · whitepaper
Under the Hood of the Variational Autoencoder (in Prose and Code)
Feb 24, 2016 · post
"Hello world" in Keras (or, Scikit-learn versus Keras)


In-depth guides to specific machine learning capabilities


Machine learning prototypes and interactive notebooks


A usable library for question answering on large datasets.

Explain BERT for Question Answering Models

Tensorflow 2.0 notebook to explain and visualize a HuggingFace BERT for Question Answering model.

NLP for Question Answering

Ongoing posts and code documenting the process of building a question answering model.

Interpretability Revisited: SHAP and LIME

Explore how to use LIME and SHAP for interpretability.


Cloudera Fast Forward is an applied machine learning reseach group.
Cloudera   Blog   Twitter