Computer Vision Engineer, Face Recognition

About the Role

You will own our face models end to end, from data pipelines and training runs to evaluation and production inference. Our models are trained in house from published research, and you will decide what we build next.

What You'll Do

• Train detection, alignment and embedding models at multi-GPU scale
• Improve accuracy on hard photos: low light, partial angles and heavy compression
• Build evaluation on public benchmarks and our own held-out sets
• Turn research papers into production code with clear ablations
• Optimize inference for latency and cost

What You Bring

• 4+ years training deep learning models for vision in production
• Strong PyTorch and distributed training experience
• Working knowledge of metric learning and margin-based losses
• Comfort reading papers and reproducing results

Nice to Have

• Published work in face recognition or detection
• Experience with large-scale data cleaning and label noise