> Markdown version of [/jobs/ext/4581801-staff-software-engineer-ml-platform-6244](https://www.wearedevelopers.com/jobs/ext/4581801-staff-software-engineer-ml-platform-6244). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# Staff Software Engineer ML Platform (6244)

- **Company:** Shield AI
- **Location:** San Mateo, CA, United States
- **Experience:** Expert
- **Salary:** $200,000 - $300,000
- **Contract:** Temporary contract
- **Skills:** Application Programming Interfaces (APIs), Distributed Systems, Python (Programming Language), Machine Learning, Azure Machine Learning, Software Engineering, Backend, Kubernetes, Free and Open-Source Software, Machine Learning Operations, Software Version Control
- **Published:** November 5, 2026
- **Apply:** https://www.dice.com/job-detail/789d1d93-6c27-4f50-a2dd-ee595638c725

## About the Role

* Experience building production backend services for machine-learning platforms or data-intensive workflows.

 * Strong software engineering skills in Python or a comparable backend language.

 * Experience designing APIs, metadata models and persistent state for services used by other engineers.

 * Experience handling storage, partial failures, retries and consistency in distributed systems.

 * Understanding of training runs, checkpoints and model versioning.

 * Ability to work directly with machine-learning users and deliver maintainable software that solves their workflow problems.

Preferred qualifications:

 * Experience with MLflow, Weights & Biases, model registries or comparable experiment-management tools.

 * Experience integrating Kubernetes jobs, object storage or machine-learning workflow systems.

 * Experience with reproducibility, data provenance or evaluation records.

 * Experience supporting open-source software or on-premises and air-gapped deployments.

## Description

Model development produces experiments, checkpoints, evaluation results and many versions of the same model. Keeping those records connected lets engineers reproduce results, compare approaches and move a trained model into use.