4 hours ago
Responsibilities
- Productionize research models through containerization, inference optimization, and deployment to edge devices and cloud GPU infrastructure.
- Build offline and online evaluation systems using production traffic and define deployment metrics such as false positives, identity switches, latency, and model drift.
- Create data feedback loops that identify failure cases, route them for labeling, and generate training and evaluation sets.
- Ship end-to-end product features from inference services and APIs through to customer-visible behavior.
- Run experiments, shadow deployments, staged fleet rollouts, and A/B comparisons between model versions.
- Manage inference cost and performance tradeoffs across constrained edge hardware and serverless GPU providers.
- Collaborate with research and platform teams to make model iteration routine and deployable.
Requirements
- Experience taking machine learning models from research code to production systems used by real users.
- Strong Python skills and working knowledge of PyTorch and modern inference runtimes such as ONNX or TensorRT.
- Experience building model evaluation pipelines, defining deployment metrics, and handling distribution shift and unlabeled production data.
- Experience with data pipelines and annotation workflows for computer vision or other perception domains.
- Comfort with Docker, Kubernetes, cloud GPU services, and CI/CD for model artifacts.
- Familiarity with computer vision, object detection, multi-object tracking, re-identification, or vision-language models.
- Experience with streaming or event-driven data systems such as Kafka or Redpanda.
- Ability to connect model improvements to customer-visible product outcomes.
- Interest in physical AI, IoT, or large-scale distributed sensing systems.
Categories
About Specter
Specter delivers real-time data and insights on private companies, enabling investors to make informed, confident decisions. Harness the power of live data and AI-driven analysis to outsmart the competition and make confident decisions in private markets.