MLOps & Data · DARVIS · 2018 — 2020

Annotation Pipeline & Team

Built the data annotation operation from scratch — tooling, team, QA and dataset versioning.

01 The problem

Engineers needed a steady supply of high-quality labelled data across regions.

02 What I did

Set up remote annotation tools on AWS and custom tooling in Python.
Led the annotation team, owned QA, storage and dataset version control.

03 Impact

Reliable labelled data for every model team