Bosch DataOps: NLP Development on Kubernetes
summary
Bosch extends DevOps into DataOps/MLOps by running a Kubernetes-based platform on AWS that supports ~30 researchers, a 5M-entity / 10M-relationship knowledge graph, and ETLs that parallelize thousands of JSON files across pods. CI/CD uses GitHub Codespaces, ephemeral Docker images, Terraform and Flux for GitOps; orchestration uses Argo Workflows with a Python Hera SDK; model lifecycle uses MLflow and Seldon with a model registry and Stardog for graph storage.
AI-written summary · about
chapters (13)
- 0:00Bosch NLP on Kubernetes
- 1:28Project goals and team
- 3:58Knowledge graph platform
- 4:40Data processing and ML workflows
- 7:10Platform requirements
- 8:37ETL workflow deep dive
- 10:21Machine learning lifecycle
- 12:05Kubernetes and tooling setup
- 13:30DataOps and MLOps implementation
- 14:57Developer workflow and GitOps
- 18:58GitOps deployment
- 21:44Future improvements
- 25:58Q&A
related talks
36mDataOps and MLOps: NLP Development and Deployment in Kubernetes made easy
29mTeam Teardown: MLOps in the Cloud for IoT
47mData Mesh in Kubernetes
52mA MLOps Platform for All Data Personas
33mScaling Open Source ML: How Wolt Uses K8s To Deliver Great Food to Mill...
18mFrom DevOps to DataOps - Cloud based Software Development
34mTransforming Data Processing with Kubernetes: Journey Towards a Self-Serve Data Mesh
29mDevOps for Data Science? - automate the boring stuff and leverage the OSS ecosystem