Containerized CV inference service that loads models and scores via API.
Shipping computer vision models often requires bespoke services per model, which slows iteration and makes deployment inconsistent across teams.
CVServe packages multiple CV models into a single containerized service with a unified API. It supports model hot‑swap, reproducible deployments, and low‑latency inference at scale.
- Docker
- FastAPI
- PyTorch
- ONNX
Phase 1
- Model loading and API‑based inference endpoints
- Standardized request/response schema
Phase 2
- Multi‑model management and versioning
- Performance tuning and scalable deployment patterns
In progress.