Everything I’ve written.

39 posts since 2018, newest first.

2026

  1. Read My PDF — turn any PDF into an audiobook
  2. Aging as an evolutionary advantage
  3. Making money
  4. Have you ever seen a failed machine learning project?
  5. Dataset drift detection is a silly idea
  6. Writing code that's breakable with tests
  7. The Art and Science of Cause and Effect — Judea Pearl
  8. Does god have human emotion?
  9. Metrics
  10. Tech debt
  11. Study
  12. Logging
  13. True randomness
  14. Grandma
  15. Thinking and calculating
  16. Best technologies
  17. Value of a test set

2024

  1. What I talk about when I talk about trees
  2. I wrote Read My PDF to read my pdf

2023

  1. On ineffectiveness of local search methods in high dimensional spaces

2022

  1. Street fighting machine learning (part 2)
  2. More on Fasttext for product embedding
  3. Be real
  4. Street fighting machine learning (part 1)
  5. Beta-Geometric model for subscription churns over time.
  6. Typical Sets and why you should not start MCMC chain from MAP
  7. Modelling changes in timeseries with sigmoids, a noodle study
  8. Match your forecast with your plan, bayesian style

2021

  1. Product embedding with fasttext, have you tried ?
  2. Quantized space SVM with entropy based sampling

2020

  1. Foolproof A/B test performance sorting
  2. Doing Kmeans IRL
  3. Quantized space SVM
  4. Predictions smedictions
  5. Rooting for Bayes

2019

  1. On kaggle public - private leaderboard
  2. The dumb blonde dashboard

2018

  1. All metrics are wrong, but some are useful
  2. Lindy's effect and how to choose the right tool.