The role
- COMP
- $160K - $240K
- EQUITY
- Competitive
- LOCATION
- San Francisco
- WORKPLACE
- Hybrid
- EXPERIENCE
- 1 - 4 years
- VISA
- None
- STACK
- PyTorch
- INDUSTRY
- AI, Hardware, Defense, Robotics, Autonomous Vehicles, Government
/roles — ROLE_466
Computer Vision Engineer
Early-stage startup building AI-powered low-light cameras for robots, drones, and defense applications
The company
Early-stage startup embedding AI into a tiny, low-power camera that sees in near-total darkness, so drones, robots, headsets, and vehicles can work at night.
- STAGE
- Series A-stage
- FUNDING
- $10M+ raised
- TEAM
- ~15 people
- FOUNDED
- 2024
- BACKING
- VC-backed
JD — the work
About the role
You would push low-light imaging forward by designing new neural architectures, training models, and running real-time inference on edge chips, drawing on your own ideas and the latest public research. Everything comes down to maximizing accuracy while keeping the model fast enough to run inside a very small camera with tight frame-rate and power budgets. The team judges ideas on whether they work, so if you have seen good ideas shut down for the wrong reasons, this may suit you.
What you'll do
- Rework the network architecture itself, with latency and image quality as your key measures
- Design and execute experiments yourself whenever you suspect a change will make the model faster or better
- Run models on edge accelerators within tight power, memory, and frame-rate limits, using SNPE, TensorRT, QNN, or AIMET
- Tackle frame interpolation, super-resolution, and denoising at the product's core, along with segmentation, optical flow, and depth
- Ship models to a major sensor maker, defense tech companies, and U.S. military branches, where results in the field decide
What they're looking for
- 1 to 4 years of experience in deep learning for computer vision
- Strong PyTorch skills for designing and training models
- Ability to take ideas from papers and your own intuition all the way to real-time inference
- Hybrid schedule based in San Francisco