Rebuilding domestic metal supply chains through hyperspectral intelligence and applied machine learning.
Spectral ML Engineer
We’re building sensing systems that can identify alloys, contaminants, and material grades in real time—directly on salvage lines and foundry floors. As an early ML hire, you’ll own the models that turn raw spectral signatures into sorting decisions, working across data collection, training pipelines, and lightweight inference on custom hardware. This isn’t ad optimization or recommendation engines; it’s gritty, physics-informed machine learning that has to run reliably in industrial environments. You’ll collaborate with a small, multidisciplinary crew in San Francisco that spans optics, metallurgy, and embedded systems, shipping models that literally determine what gets melted down and reused.
What they're looking for
- Fluency in Python and PyTorch; comfortable owning an ML pipeline from data ingestion through deployment
- Experience with spectral data, signal processing, or any sensor modality where noise and calibration are first-class problems
- Instinct for working with messy, real-world data—missing scans, mislabeled samples, sensor drift—and engineering solutions that hold up in production
- Strong communication habits and a bias toward rapid prototyping paired with clear documentation
- 0–4 years of relevant experience, whether from industry, research, or hands-on projects outside a traditional ML setting