Founding Engineer.

About Seawolf

Seawolf Technologies is replacing helicopter spotting and spotter planes with AI. We use computer vision on commercial drones to detect tuna from the air, giving fishing captains a faster, cheaper, and more reliable way to find fish. Fishing is one of the last major industries to be digitized, and we're here to change that.

What you'd work on

You'd own the detection system alongside the founding team: designing the model architecture, running experiments, and iterating toward a production system that holds up against salt spray, harsh lighting, and constant motion. You'd have a direct hand in product and technical strategy. Engineering decisions here are company decisions.

If you've ever wanted your ML work to involve the Pacific Ocean instead of another dashboard, this is it.

What we're looking for

You've trained object detection models on custom datasets in PyTorch, with real data that fought you: small objects, bad annotations, domain shift, models that pass on the test set but fail in the field. You know the YOLO and DETR families deeply enough to modify an architecture, not just configure one, and you can build a detection pipeline from open-source components rather than calling a high-level API.

You form your own opinions about how ML systems should be built, and you say them out loud. You'll own the vision system, but you're the kind of engineer who reads the pipeline code, pushes a fix, and has a point of view on how the training infrastructure should work.

Preferred

• Aerial imagery, remote sensing, or geospatial experience.
• Deploying and optimizing computer vision models for real-time inference on edge devices, including profiling, quantization, and balancing accuracy, latency, memory, and power constraints.
• Wildlife or environmental detection.
• Prior startup experience.