Role OverviewWe're seeking a
research engineer to bridge the gap between research and production, building
ML infrastructure and
data systems for medical imaging at scale. You'll own critical data pipelines that unify live production traffic with offline datasets, design storage solutions for multimodal medical data, and build training + inference infrastructure that enables our research team to iterate rapidly. This role requires someone who can move fluidly between
model training, data engineering, ML systems, and
production deployment.
Key Responsibilities- Build and optimize distributed ML infrastructure for training foundation models on large-scale medical imaging datasets.
- Design and implement robust data pipelines to collect, process, and store large-scale multimodal medical imaging data from both production traffic and offline sources.
- Build centralized data storage solutions with standardized formats (e.g., protobufs) that enable efficient retrieval and training across the organization.
- Create model inference pipelines and evaluation frameworks that work seamlessly across research experimentation and production deployment.
- Collaborate with researchers to rapidly prototype new ideas and translate them into production-ready code.
- Own end-to-end delivery of ML systems from experimentation through deployment and monitoring.
Qualifications- 5+ years building ML infrastructure, data pipelines, or ML systems in production
- Strong Python skills and expertise in PyTorch or JAX
- Hands-on experience with data pipeline technologies (e.g., Spark, Airflow, BigQuery, Snowflake, Databricks, Chalk) and schema design
- Experience with distributed systems, cloud infrastructure (AWS/GCP), and containerization (Docker/Kubernetes)
- Track record of building scalable data systems and shipping production ML infrastructure
- Ability to move quickly and handle competing priorities in a fast-paced environment
Preferred Qualifications- Experience with reinforcement learning training pipelines (e.g., RLHF, reward modeling, or online learning systems)
- Support A/B testing and experimentation workflows for model rollouts, including monitoring statistical significance and managing canary deployments.
- Familiarity with vision-language models (VLMs) or multimodal architectures
- Experience with medical imaging formats (DICOM) and healthcare data standards
- Background in distributed training frameworks (PyTorch Lightning, DeepSpeed, Accelerate)
- Familiarity with MLOps practices and model deployment pipelines
- Experience with privacy-preserving data systems and HIPAA compliance
