Building a Legal Data Service with Clojure
summary
Cicayda builds a multi-tenant e‑discovery service in Clojure/ClojureScript that ingests hundreds of millions of facts at once, uses Datomic as the authoritative store and hits its transactor limits early, so job tracking moves out of Datomic and a central ingestion pipeline with SQS-like queues, a resource multiplexer and a declarative routing slip coordinates backpressure. The result cuts transactor unavailable errors to only true outages, enables safe large migrations and lets the UI use DataScript and core.async for smoother interactions.
AI-written summary · about
chapters (18)
- 0:00Intro
- 0:35eDiscovery
- 4:11Proposition 1
- 7:35Our Environment
- 9:01Datomic
- 10:44How'd That Work Out?
- 12:45WHAT DO I CHOOSE?
- 13:28Missing Decisions
- 16:11Decisions Made
- 20:30Design Payoffs
- 24:58Digging for Constraints
- 25:44UI Overview
- 27:10Libraries We Use
- 27:43Library Upgrades
- 30:16Useful Upgrades?
- 32:40Library Driven Development
- 33:36Design Driven Development
- 36:36Influences
related talks
33mClojure Dispatches From a National Research Lab
43mDoing Data Science with Clojure
38mUnlocking data-driven systems
13mArchitecting Threat Intelligence with Clojure in Cybersecurity Detection & Response
44mRaising Services and People with Clojure
28mDatomic at Nubank, Simplicity Scaled
46mClojure Scaling the Event Stream
37mClojure Remixed: Pyroclast Data Processing