21 hours ago
Base Salary
$200k - $260k/yr
Responsibilities
- Design computer vision architectures for fine-grained item identification and adapt foundation vision models to the company’s catalog.
- Define category-level accuracy standards and build calibrated confidence scoring for communicating model uncertainty.
- Create model-error feedback loops to guide labeling priorities with the labeling team.
- Prioritize category expansion versus accuracy improvements in existing categories.
- Own the model serving path from research through production, balancing latency, cost, and reliability at scale.
- Explain model capabilities, limitations, and technical tradeoffs to stakeholders and lead technical direction on an open-ended problem.
Requirements
- At least 5 years of applied computer vision experience, including shipping production systems at meaningful scale.
- Hands-on experience with fine-grained or instance-level recognition of visually similar products.
- Production experience fine-tuning and deploying vision transformers or CNNs using PyTorch or TensorFlow.
- Experience adapting foundation vision models to specific use cases and designing evaluation frameworks beyond loss metrics.
- Experience with calibrated prediction confidence and model-serving decisions involving latency, cost, and reliability.
- Clear written and verbal communication skills and comfort leading technical direction.
- Experience with active learning, human-in-the-loop labeling, low-latency APIs, or early-stage startups is useful.
Benefits
- Hybrid role based in New York, New York.
- Visa sponsorship is not available.
Tech Stack
PyTorchTensorFlow