80% less manual work on AxiomSL rule updates
Two tools that put language models to work on data operations: tracing incidents and updating rules.
Context
When a report breaks, someone has to trace the data back through several systems by hand. At RBC, AxiomSL rules were also updated by hand.
What I did
I built an OpenLineage-based lineage tool that documents multi-source data flows automatically and uses an LLM to suggest likely root causes. I also automated AxiomSL rule updates at RBC with an LLM, keeping governance controls in place.
Outcome
80% less manual work on AxiomSL rule updates. Shorter investigations across about three incidents a month.
Evidence
Client work, no public page.
Stack
Python, OpenLineage, LLMs, AxiomSL