The role
- COMP
- $250K - $300K
- EQUITY
- Competitive equity
- LOCATION
- San Francisco
- WORKPLACE
- On-site
- EXPERIENCE
- 6+ years
- VISA
- None, Visa transfers, New visa sponsorships
- STACK
- Python, PyTorch, Kubernetes, K8s, Docker
- INDUSTRY
- Healthcare, AI
/roles — ROLE_189
Software Engineer - ML Infrastructure
Small, well-backed team training foundation models on medical scans to ease a critical radiologist shortage
The company
An early-stage medical AI startup pairing radiologists with its own multimodal imaging copilot to deliver radiology reads as a service, aimed at a severe shortage of radiologists.
- STAGE
- seed-stage
- FUNDING
- seed funding
- TEAM
- founding team of <10
- FOUNDED
- 2023
JD — the work
About the role
You would come in as a senior, self-directed engineer to define how the team trains models across many machines, runs reinforcement learning, and serves inference, and to own the systems behind the company's foundation models for medical imaging. The ideal person has run ML infrastructure at scale inside a top engineering organization and can show a small team what excellent looks like. It is a ground-floor role on a team of about five, backed by respected investors, building a radiology service around its own multimodal AI copilot.
What you'll do
- Own training infrastructure for imaging foundation models across many GPUs, handling parallelism and checkpoints for 3D scans
- Build the reinforcement learning stack, from high-volume rollouts to serving reward models and gathering experience data
- Collaborate with researchers from leading AI labs, turning their experiments and prototypes into production systems
- Create data loaders and preprocessing that keep accelerators fully fed with large 3D and multimodal imaging data
- Help run model deployment and serving, with staged rollouts, canaries, and monitoring
What they're looking for
- 6+ years of experience, including running ML infrastructure for large workloads at a leading tech company
- Deep knowledge of RL systems, large-scale distributed training, and model inference
- Ability to work autonomously and teach a small team strong ML infrastructure practices
- Hands-on experience with Python, PyTorch, Kubernetes, and Docker