ISSUE 347 · August 3, 2021
In the News
AI doesn’t have to be too complicated or expensive
For companies that are interested in using AI but don't have access to vast troves of data, there hasn't been a clear path to follow. But because of that, those industries may be the best untapped opportunities for AI. In this article, Andrew Ng takes a look at the issues, the opportunities, and why now is the time to get started.
Harvard Business Review | Andrew Ng
Data Science & Analytics Salary Increases in 2021
This summary of a new Burtch Works Report shows how salaries have changed over the past year and what to expect going forward. For data science and analytics professionals, the trends look great. Follow the links to download the full 50 page report.
Burtch Works
Sponsored Link
Add Vector-Search to Production Applications
Pinecone makes it easy to add vector similarity search to production applications. No more hassles of tuning algorithms or building and maintaining infrastructure. Try it for semantic text search, image/audio search, recommendation systems, and other applications.
Tutorials, Projects & Opinions
What Have Language Models Learned?
Great interactive post that shows how language models understand the world. Especially, if language models are new for you, this is a must-read post!
Google | PAIR Explorables
What's bad about Julia?
This post is about all the major disadvantages of Julia, written by an knowledgeable Julia user and fan. "Learning why you may not want to choose to use a tool is just as important as learning why you may."
Jakob Nybo Nissen
SQL Snippets
If you have a notebook with miscellaneous bits of SQL code, you'll appreciate this searchable, crowd-sourced collection of SQL snippets. Covers a variety of uses for PostgreSQL, BigQuery, Redshift and more.
Count
Share Databases, Not Files
Help eliminate the CSV. bit.io is the fastest way to create, share, and collaborate on a real database. Get a Postgres-compatible database in one-click with easy, secure permissions and zero management. Join us in creating a single place for public and private data that makes everyone immediately productive with data. Try it now!
// sponsored
Code & Tools
QuestDB
QuestDB is an open-source SQL database that's built for performance and optimized for time-series data. It's currently used in production environments and claims to be the fastest open-source option for time-series. Follow the links for a live demo and there's also a worthwhile discussion on Hacker News about its capabilities.
GitHub | QuestDB
Shapash
Shapash is a Python library that provides a variety of visualizations to help make machine learning models easy to interpret and understand.
GitHub | MAIF
Resources
Awesome MLOps
This curated collection of MLOps resources includes books, articles, papers, talks, tooling, newsletters, communities and more. Start with the first section, "MLOps Core."
Larysa Visengeriyeva
Outlier
The Jessica Simulation
After the death of the woman he loved, Joshua Barbeau turned to a GPT-3 powered website to speak to her once again. These are early days for experiments like this but gven enough journal writings, letters, posts and tweets, could a GPT-3 bot be used to make someone live forever? And if so, would that be a good thing?
San Francisco Chronicle | Jason Fagone
 |