Senior Machine Learning Engineer, Agentic Science/Generative Models, AI for Biology & Translation (AIBT)
Roche3 days ago
Base Salary
$168k - $312k/yr
Responsibilities
- Build agents that use tools, retrieve evidence, and reason across multi-step scientific workflows.
- Develop interfaces between agents and biological, genomic, and clinical data sources.
- Design evaluation harnesses that assess agent outputs against scientific ground truth.
- Implement self-improving and autonomous loops for AutoML and lab-in-the-loop workflows.
- Build agent memory and context-management systems for long-horizon workflows.
- Build, fine-tune, deploy, and scale foundation models and LLMs in production.
- Own production Python/PyTorch or JAX codebases that turn research ideas into reliable, reusable software.
- Own the MLOps and AgentOps lifecycle, including experiment tracking, evaluation, monitoring, reproducibility, and infrastructure-as-code.
- Work with research scientists to turn open-ended scientific problems into scoped, shippable systems.
- Lead technical projects end to end and help raise the engineering bar across the organization.
Requirements
- BS or MS in computer science, machine learning, engineering, or a related quantitative field.
- At least 5 years of experience building and shipping ML systems in industry.
- Excellent Python skills and strong software and data engineering fundamentals.
- Experience with Git, automated testing, CI/CD, and documentation.
- A track record of leading technical projects end to end.
- Comfort with ambiguity and close collaboration with scientists.
- Strong problem-solving and communication skills.
- Interest or experience applying ML to scientific discovery, including biology, chemistry, drug discovery, domain-specific data, or domain-specific models.
- Preferred: experience with inference-time scaling and optimization, including test-time compute, sampling and search, model routing, batching, caching, and latency/cost/quality tradeoffs.
- Preferred: ML infrastructure on AWS, including EC2, S3, EKS, SageMaker, distributed training, and inference on HPC.
- Preferred: evaluation systems for agentic applications with scientifically defined correctness.
- Preferred: production agent orchestration frameworks such as LangGraph and MCP-based tool integration.
Benefits
- The position qualifies for the company's benefits package.
- Relocation benefits are not available for this job posting.
- The expected hiring range is based on the primary location of San Francisco; actual pay depends on experience, qualifications, geographic location, and other permitted factors.
- A discretionary annual bonus may be available based on individual and company performance.
Categories
About Roche
Roche is a Swiss healthcare company that develops and sells prescription pharmaceuticals and in vitro diagnostic systems for hospitals, laboratories, and clinicians. Its business spans two core divisions—Pharmaceuticals and Diagnostics—with strengths in oncology, immunology, neuroscience, infectious diseases, and diabetes care. Founded in 1896 and headquartered in Basel, Switzerland, Roche also advances personalized healthcare by pairing therapies with companion diagnostics.