Search relevance tuning for a subscription box service
Search relevance tuning for a subscription box service. Owned the model, the evaluation and the deployment. Click-through on recommended items improved by 31%.
Loading the home page
Daichi Sato
Machine Learning Architect | Recsys & Ranking at Scale
Rate
$110.00/hr
JPNot taking work0 contracts completedNo reviews yet
About
Recommendation systems are what I do. I led ranking work at a Tokyo media company serving 8M monthly users, where a reranking model I shipped lifted session depth by 11%. These days I take on one client at a time, usually helping teams move from collaborative filtering hacks to something they can retrain nightly without fear.
Portfolio
03 filedDocument question-answering for a subscription box service
Document question-answering for a subscription box service. Owned the model, the evaluation and the deployment. First-response time on support tickets dropped from hours to minutes.
Support triage classifier for a music streaming service
Support triage classifier for a music streaming service. Owned the model, the evaluation and the deployment. First-response time on support tickets dropped from hours to minutes.
Reviews
No reviews in print yet
Feedback stays sealed until both sides of a completed contract have written theirs, then both are published at once. Either no contract has finished here yet, or one side is still holding its review.