ML is cool. Here's a Github repo with a bunch of detailed paper notes, reviews, and summaries.

Most of these were read and reviewed with Paper Club, a group of 5-6 intrepid learners that I started with my friends James Vanneman and Tiger Shen in April 2017 after we had finished all the courses at the Bradfield School of Computer Science.

My learning journey so far:
  • summer '16: organized a coworkers' book club for Introduction to Statistical Learning in R
  • spring-summer '17: part 1 & 2
  • fall '17: reading papers and writing blog posts on our Medium publication
  • winter '17: NIPS '17 field trip, implementing papers, hired as contract Machine Learning Engineer at Sourceress
  • spring '18: diving deep on Bayesian methods
  • also spring '18: hired full-time as Founding MLE at Sourceress
  • summer-fall '18: working through a course I've organized called Machine Learning Engineering in 10 Weeks
  • winter '18/19: organizing group programming days at the Archive, implementing papers and working on fun projects
  • spring '19: facilitating a second run of MLE in 10 Weeks. From a review from one participant: "I feel like a much more confident and prepared machine learning practitioner with these tools in my toolbox."
  • also spring '19: build the creativity-stimulating Duet, enabling users to write collaboratively with OpenAI's state-of-the-art language model (with Austin Hou and Tina Kim)

This field is great fun - it moves very quickly, you can have an impact on many different and otherwise unrelated fields (because it's very general), and it's still pretty rare. I remember thinking in 2016 that it was too late to get into machine learning, and now I realize more than ever that it's still early days. Another thing that makes it fun is that it's young and still relatively shallow: unlike physics where it takes a 5 year PhD to make a small improvement, you can reach the cutting edge of an ML subfield within a year or two of learning (assuming a programming background), and PhDs are often not necessary even for research positions.

Here are some more details about the materials I found most helpful.
Updated: August 13th, 2019

Sources, notes:

Key: ✍️ is a blog post, 📘 is a flashcard, no emoji means notes & highlights.
Rethinking S3: Announcing T4, a team data hub – Quilt October 22nd, 2018
Hardware for Deep Learning. Part 1: Introduction – Intento September 27th, 2018
Efficient and Robust Automated Machine Learning (auto-sklearn paper) September 17th, 2018
✍️ How I learned web development, software engineering, & ML August 4th, 2018
✍️ Learning to be a power user of operating systems July 26th, 2018
High-Skilled White-Collar Work? Machines Can Do That, Too July 8th, 2018
✍️ Machine Learning Engineering in 10 Weeks curriculum v1 July 7th, 2018
Rebooting AI - postulates July 5th, 2018
How not to structure your database-backed web applications: a study of performance bugs in the wild July 4th, 2018