Kaur Jaanson

Joined via Zendesk's acquisition of Klaus in February 2024 and contributed to migrating Klaus' data pipelines and ML services from GCP to Zendesk's AWS/Kubernetes platform. Drove the prototyping and initial rollout of moving Apache Beam pipelines off managed Dataflow onto a self-hosted Flink cluster on Kubernetes — discovery, proof of concept, and first pipelines into production. Swapping the Beam runner rather than rewriting the pipelines kept existing business logic intact and reduced the migration to an engineering problem. Also took part in the subsequent optimization, in porting GCP Cloud Function ML services to Kubernetes, and in planning the wider migration to AWS and Zendesk tooling.

ML engineer at Klaus, an AI-powered quality management platform for customer support, working on models and data pipelines for automated QA of customer interactions.

Working in Microsoft IC3 Fraud Data Science team to analyse and develop metrics for fraud, abuse and spam pattern analysis. Designed and setup of data-science backend environments that would allow team members to conduct analysis and prototype machine learning solutions. Productionized initial ML model prototypes created by data-scientists. Working on internal compliance reviews to make sure that teams’ data-science solutions would be compliant with company's privacy and security standards.