Onyx-Native, Information Flow Models and Scientific computing
summary
Onyx-Native is a library that injects stateful, native-code–backed Java objects into an Onyx Platform workflow so polyglot legacy pipelines become declarative information flows. Applied to the ASOS Ingest System (a 15‑year production pipeline with 12,605 lines of bash and Java/C/C++/Fortran glue), it uses a Java API plus JNI/Fortran interop to bootstrap native libraries, preserves existing file formats, and cuts an expected 2–4 year rewrite down to months while keeping scientific correctness.
AI-written summary · about
chapters (8)
- 0:00Start
- 1:16Scientific Computing
- 10:24ASOS: System Diagram
- 12:43Human-Scale Stabilities
- 15:18Information Flow Model (Highlight Reel)
- 31:54Unexpected Payoffs
- 35:22Bold Claims!
- 43:18The Onyx Platform Catalog of Tasks
related talks
21mDesigning Modular Data Pipeline
49mWorking on Cancer
33mAI Context Engineering for R&D Knowledge Using a Multi-Model Database
04mInterview
43mChange data capture in production with Apache Flink
50mGetting started with MicroStream - High-Performance Java-Native Persistence
35mGame Development Development
52mData Over Code: A Short Hyperbole