Architecture Decisions: A Case Study -
summary
Prototype swaps Node.js for Elixir and Erlang after a one-week spike, rebuilding most of the platform in a month; the codebase initially contains 70% accidental complexity so the team runs import, analysis and UI in memory to iterate fast on a sample of ~40,000 medical events. They expose LiveView + Tailwind demos on Heroku, introduce a tiny DSL then drop it, separate import/analysis/UI, add batched Postgres event persistence, and achieve rapid, working clinical feedback.
AI-written summary · about
chapters (10)
- 0:00Introduction to startup ADRs
- 3:20The medical project context
- 7:13Early technical decisions
- 11:27Data handling and processing
- 15:44Domain modeling and language
- 20:50Evolving the system structure
- 25:28UI and cloud deployment
- 30:12Separating concerns and state
- 37:10Refining rules and persistence
- 42:06Retrospective and core insights
related talks
36m"Engineering to embrace change" Or "How clumsy engineering can lead to powerful research – Part II"
47mElixir for curing product development pains
47mElixir for curing product development pains?
39mKeeping Code Consistent
57mMaximising Side Effects When Adopting Functional Programming
06mResilient Code, Resilient Engineer: Lessons from Changing Requirements
37mModern Abstractions and the Web, and avoiding Pitfalls because Life is short
01h 02mFishbowl @ Functional Conf