23 CDN-sharded data repositories keep lookups scoped instead of shipping the complete dataset to every browser.
Kiokun
A unified Chinese–Japanese dictionary.
Full-stack · Data engineering · Product design
Problem
Chinese and Japanese dictionaries expose overlapping character systems as separate products. I wanted one fast lookup that could reconcile both languages without shipping a giant database to every browser.
What I built
I built the web product and the data pipeline behind it: ingesting and normalizing multiple dictionary sources, aligning variants with OpenCC, generating the final entries, and distributing the result as CDN-sharded JSON so the browser only loads what it needs.
Evidence
A Rust pipeline ingests and normalizes multiple dictionary sources before the site build.
OpenCC variant alignment connects Chinese and Japanese forms inside one entry model.
Live product