Machine Learning Operations Engineer

Our company develops advanced software for biopharmaceutical process development and manufacturing. Our platform combines AI, mechanistic modeling, and bioprocess expertise to build predictive digital twins that support process understanding, optimization, and better decision-making. We work with leading biopharmaceutical companies and are building a high-quality SaaS platform with a strong focus on usability, reliability, maintainability, and scientific value.

We are seeking a Machine Learning Operations Engineer to build and standardize the ML platform behind our biopharma process modeling products. This is a platform and tooling role: your job is to make ML development easier, safer, and more reproducible for Process Modeling Scientists and Machine Learning Engineers.

You will own the systems around ML — workflows, artifacts, experiment tracking, dataflow, lineage, reproducibility, governance, and delivery. This is not primarily a model-research role.

In our domain, model quality is not only accuracy, loss, or inference speed. A pipeline or platform change can look fine on standard ML metrics and still be unacceptable if it alters scientific outputs such as metabolite, salt, or process-variable predictions, or breaks biological constraints. Protecting reproducibility, traceability, and scientific validity across the ML lifecycle is central to this role.

• Own ML workflow orchestration using Flyte or similar systems.
• Automate repetitive experimentation and delivery tasks to reduce manual work.
• Improve the reliability and debuggability of ML workflow execution.

• Own experiment tracking using MLflow or equivalent tools.
• Manage artifacts for datasets, models, reports, validation outputs, and scientific results.
• Design dataflow and artifact flow across the ML lifecycle.

• Ensure versioning, lineage, traceability, and reproducibility of ML workflows.
• Build compatibility and validation checks that prevent unsafe reuse of models or artifacts.
• Define standards and best practices for scientific ML governance.

• Build observability into workflow execution: structured logs, metrics, run state, and failure reasons.
• Design safe recovery, rollback, and rerun paths for failed or suspicious runs.
• Turn recurring problems into reusable platform improvements.

• Strong software engineering background with strong Python skills.
• Practical experience with MLOps, ML platforms, workflow orchestration, experiment tracking, or artifact management.
• Experience with Docker and Kubernetes or similar systems.
• Experience with cloud environments such as AWS, GCP, or Azure.
• Good understanding of CI/CD, reproducibility, observability, and production-grade software practices.
• Ability to work with complex data workflows and distributed systems.
• High autonomy: notices what is broken, fixes it, and turns repeated problems into reusable platform improvements.
• Ability to work remotely with at least 4 hours of overlap with CET working hours.
• B.S. in Engineering, Computer Science, Data Engineering, or a related technical field.

• Experience with Flyte, Air Flow or similar systems.
• Experience building internal developer platforms or ML platforms.
• Experience with data lineage, artifact versioning, or model registry practices.
• Experience with scientific computing or industrial ML workflows.
• Familiarity with model training, evaluation, and deployment workflows.
• Experience working with scientists or domain experts.

• Strong ownership mindset with genuine interest in basic science applied to industry.
• Clear communication across engineers, scientists, and domain experts.
• Pragmatic and reliability-focused, balancing governance and rigor with delivery speed.
• Structured, detail-oriented approach to reproducibility and traceability.
• Comfortable working independently in a fast-moving, cross-functional environment.

You will help shape the ML platform behind a scientific SaaS product, working with scientists, engineers, and domain experts on real biopharma process development and manufacturing problems. You will build the tools and standards that let ML work move faster without losing reproducibility, traceability, or scientific context.