Inference Engineer
moondream · San Francisco, CA
Frontier labonsite$165k–$215k
At a glance
Highlights
- Own Moondream's inference performance end-to-end from profiling to deployment.
- Make vision models run in realtime over every frame of video.
- Partner with research so new architectures are fast by design.
What they offer
- Meaningful equity
- Health, dental, and vision
- Paid parental leave
- Relocation support
Ideal candidate
You're a performance-obsessed engineer who loves squeezing speed from real models on real hardware and supporting them in production. You think in bottlenecks and tradeoffs, and enjoy collaborating in-person to build fast systems.
Requirements
- Made production inference workloads meaningfully faster
- Written shipped CUDA kernels and can reason about bottlenecks
- Used TensorRT, Core ML, XLA, or Neuron SDK in production
- Have opinions about when to fuse kernels and when not to
- Shipped and supported software on hardware you didn't choose
- remote policy
- In-person in San Francisco; no remote work
- hiring process
- Send your resume to hiring@moondream.ai; expect to hear back within a week
Lab profile
About moondream
Builds fast, accurate vision AI for realtime understanding of every video frame
What they build
- Realtime vision AI models
- Video inference APIs for developers
Known for
- Accurate yet fast vision models
- Realtime every-frame video processing
- Edge and on-premise deployment
- Highly optimized inference stack
Posted · Last seen