Using Elixir and WhatsApp to Response COVID-19
summary
Built a WhatsApp COVID-19 hotline in five days using Elixir + Phoenix on Kubernetes (AWS), Postgres, factory job workers and the WhatsApp Business API, with QR/bitly round‑robin routing; load tested to 1,800 req/s with loader.io. The South African rollout hits ~8 million subscribers (450k on day one prelaunch) and the WHO service reaches 10 million users in 48 hours. Elasticsearch is disabled at launch due to ops complexity and some Terraform/automation shortcuts create ongoing technical debt.
AI-written summary · about
chapters (21)
- 0:00Intro
- 1:08Strategy
- 2:37Reminders
- 3:31Mixed Media
- 4:03Behaviors
- 4:40Reminder
- 6:08Covid Connect
- 6:43The WHO
- 7:52Use Case
- 8:24Latest News
- 9:05Launch Day
- 10:13Timeline
- 13:51Stateful design
- 14:46The stack
- 16:10Phoenix web framework
- 17:58Stack overview
- 19:53Things that worked
- 26:07Things that didnt work
- 27:59Learnings
- 34:17Questions
- 41:03Risks
related talks
44mIdioms for Building Fault-tolerant Applications with Elixir
49mAn Introduction to Erlang From behind the Trenches
38mElixir Adoption at Scale: A Comprehensive Guide
41mIdioms for building fault-tolerant applications with Elixir
44mMedical Messaging: An Elixir For Healthcare
53mIdioms for Building Distributed Fault-tolerant Applications with Elixir
03h 19mWorkshop - All Aboard The Elixir Express!
01h 52mPurple Carpet