Machine Learning Engineer, Computer Vision
Object detection and tracking for deployed systems, from a paper to a model running at the edge.
What you would do
- Build and improve detection and multi-object tracking models on customer footage, and measure them honestly against the metrics the customer cares about.
- Take a method from a recent paper and implement it well enough to know whether it helps.
- Export, quantize and profile models for ONNX Runtime and NVIDIA Jetson targets, and own the gap between the training number and the deployed number.
- Work directly with the customer's engineers on billable engagements, with the documentation that implies.
What we screen for
The screening exercise measures exactly these three things.
- Detection or tracking work that went past a tutorial: your own data, your own failure cases, your own fixes.
- Comfort with ONNX or another edge deployment path, including the things that break during export.
- Reading a short method description and turning it into working code.
Requirements
- New graduate to roughly two years of applied experience in computer vision or machine learning engineering.
- Authorized to work in the United States. Some of the work this role supports requires United States person status.
- Comfortable being the engineer a customer talks to.
Questions or accommodations: careers@di-ds.com. DI-DS is an equal opportunity employer.