Canva Engineering Blog
canvatechblog.com · Company blog · on HN since 2021
Canva's engineering team writes about scaling media, search, real-time collaboration, and GPU machine learning infrastructure.
What HN saysHN commenters bring it up in discussions of GPU infrastructure on Kubernetes, and an author appeared in comments to correct a point about translation files.
17HN people shared it
2linked it in comments
54stories submitted
730story points
4 years agolast shared
2021people per year2026
Most-shared posts
- 1Supporting GPU-Accelerated Machine Learning with Kubernetes and Nix4 people · 11 points · first shared 2022 · discussion on HN
- 2From Zero to 50M Uploads per Day: Scaling Media at Canva4 people · 6 points · first shared 2022 · discussion on HN
- 3Simple, Fast, and Scalable Reverse Image Search2 people · 131 points · first shared 2022 · 31 comments on HN
- 4Hyperparameter Optimization with Argo2 people · 30 points · first shared 2021 · 6 comments on HN
- 5Enabling real-time collaboration with RSocket2 people · 14 points · first shared 2021 · 2 comments on HN
- 6Subscription Analytics at Scale2 people · 4 points · first shared 2021 · discussion on HN
- 7Deep Learning for Infinite (Multi-Lingual) Keywords2 people · 2 points · first shared 2023 · discussion on HN
- 8We Put Half a Million Files in One Git Repository, Here’s What We Learned1 people · 222 points · first shared 2022 · 257 comments on HN
- 9CSS: Absolutely positioning things relatively1 people · 134 points · first shared 2022 · 38 comments on HN
- 10How to design in every language at once1 people · 85 points · first shared 2021 · 25 comments on HN
- 11Combining Classic and Modern: A New Approach to Camera Simulation1 people · 18 points · first shared 2021 · 10 comments on HN
- 12Search Pipeline: Part I1 people · 16 points · first shared 2022 · discussion on HN
- 13Migrating from Solr to Elasticsearch, and their differences1 people · 10 points · first shared 2022 · 4 comments on HN
- 14Testing SEO scalably without breaking a sweat1 people · 5 points · first shared 2022 · discussion on HN
- 15Recommender systems: When they fail, who are you gonna call?1 people · 4 points · first shared 2021 · discussion on HN
What HN people wrote when they linked it
thundergolfer · 2023-04-04
Just use https://modal.com/ :)
At Canva I built auto-scaling GPU infra on K8s for model training[1], and it's way too much work and operational expense to be worth building yourself. I went work at Modal because building it properly once and then distributing the solution was going to be just way better and more efficient.
1. https://canvatechblog.com/supporting-gpu-accelerated-machine-learning-…
haizzz · 2022-06-16
Hey everyone, author here, the article is a bit misleading in that .xlf files aren't really generated files, they're created through our translation pipeline by real humans. I considered them generated in the sense that they're not directly worked on by engineers who have to deal with them in the repository.
The content of these translation files are snapshot in time aligned with the text in our product so simply removing them we would lose all the changes made to translations each time texts are changed.
Sorry about that, hope this clears it up!
Edit: for more information on translation and xlf files, we have another blog post all about them
https://canvatechblog.com/how-to-design-in-every-language-at-once-f2dd…