Data & MLOps
Production Inference Platform
Deepecho
The serving layer that puts the ultrasound models into production: FastAPI in front of Triton, ONNX models on cloud GPUs.
- →Client → FastAPI → Triton → GPU inference → results
- →Model export to ONNX and OpenVINO; batch and online inference
- →Containerized on Google Cloud with PostgreSQL for results
FastAPITritonONNXOpenVINODockerGCPPostgreSQL