|
|
December 23 · Issue #33 · View online
Curated essays about the future of Data Science. Production Data Science and learning resources for continuous learning. Covers Data Science, Data Engineering, MLOps & DataOps. Curated by people at https://nibble.ai/
|
|
đ Want to support this project? Forward this email to a friend! Hello dear reader, This is our last issue of 2019! While the 2010s have seen the rise (and hype) of applied data science, we hope the next ten years will be about bringing it to maturity. Whatâs next? nibble.ai dispatch will resume in 2020. What started as a weekly release is shifting to a more steady pace. While the fast release cycle was perfect for quick iteration, we now want to make sure we can maintain quality while working on other projects.
Our editorial line will stay the same:Â to help you become a better data practitioner and be prepared for the future of data science. This takes the form of curating relevant industry essays, news about tools and learning resources that will sharpen your technical abilities.
Besides this publication and client work, weâve been working on an exciting new project: a series of MLOps events in Paris. This aim at bringing more awareness to the field and help practitioners learn from each other. We will be launching the series with Managing the Machine Learning lifecycle with MLflow next January, 16th.
I had a great time preparing these issues, and I hope you enjoyed reading them. Wishing you the best for 2020, Florent
|
|
Wizards of Oz?
|
AI update, late 2019 â wizards of Oz
The editorial line of Filip Piekniewskiâs blog is about âstripping the AI news coverage out of fluff and trying to get to the substance, often with a fair dose of sarcasm and cynicismâ. This post does exactly this. The whole field of AI resembles a giant collective of wizards of Oz. A lot of effort is put into convincing gullible public that AI is magic, where in fact it is really just a bunch of smoke and mirrors.
|
Adapt or Die: why your business strategy is failing your data strategy
Data projects are failing, but the problem isnât your data team â find out why business goals and data outcomes arenât lining up and what to do about it. Because of the difficulty of working with data, weâre seeing a growing dissatisfaction with data projects. The hype is there, the talent is there, but the results arenât. Months or even years of work are going into ideas that arenât being productionalized due to the operational hurdles involved in using data effectively.
|
MLOps â Data Skeptic đ
Kyle meets up with Damian Brady at Microsoft Ignite 2019 to discuss MLOps. Damian is a DevOps specialist and recently started spending a lof of time in the world of machine learning and data science.
tl;dr: Software development has grown quite a lot from its early days, but machine learning hasnât quite caught up yet. The conversation explores ways machine learning development can grow in maturity (artefacts versioning, deployment, data drift) and tools that can enable it.
|
Enterprise Readiness, MLOps and Lifecycle Management đ
In this episode, Sam is joined by Jordan Edwards, Principal Program Manager for MLOps on Azure ML at Microsoft, to discuss MLOps and lifecycle management.
tl;dr: A broad conversation ranging from testing in an MLOps environment, managing model lifecycles, the stages along the journey of implementing MLOps, how companies should look at hiring ML Engineers vs DevOps EngineersâŚ
|
|
Netflix open-sources Metaflow
A pipeline framework for data science. While there are already existing frameworks, like Airflow or Luigi, which allow the execution of DAGs consisting of arbitrary Python code, Metaflow offers a different approach which is perhaps more suited for data science work. The library is available on github.
|
TensorFlow introduces TensorBoard.dev
Only available in preview mode at the moment, TensorFlow has just introduced TensorBoard.dev which aims at facilitating collaboration. We have seen people sharing screenshots of their TensorBoards to achieve this. However, screenshots arenât interactive and fail to capture all the details.
|
|
Machine Learning Systems design
A booklet on machine learning systems design with exercises. When searching for a solution, your goal isnât to show off your knowledge of the latest buzzwords but to use the simplest solution that can do the job.
|
Monoid in the Category of Endofunctors
A deep dive into the meaning of the famous sentence âMonoid in the Category of Endofunctorsâ which is actually the definition of a Monad. Iâve always liked functional programming and plan to get deeper into it next year, in part for my own intellectual satisfaction, but also because the functional paradigm really makes sense for many things involving data.
|
|
Hope you liked this issue. If you did, make sure to give it a thumbs up đ!
|
Did you enjoy this issue?
|
|
|
|
In order to unsubscribe, click here.
If you were forwarded this newsletter and you like it, you can subscribe here.
|
|
|