Note: The job is a remote job and is open to candidates in USA. Pictor Labs is the leading virtual staining company revolutionizing digital pathology adoption worldwide through cutting-edge AI-powered technology. They are seeking an experienced Senior ML Inference Engineer to join their team, focusing on optimizing and deploying production virtual staining models at scale.
Responsibilities
- Design, development, and optimization of production ML inference systems for virtual staining models (Deepstain, Restain, ClearStain) serving clinical and pharmaceutical customers
- Architect and implement high-performance inference pipelines capable of processing gigapixel pathology images with sub-2-minute latency requirements
- Work with ML Research and Engineering teams to optimize model architectures and deployment strategies for both cloud-based APIs and edge devices (NVIDIA DGX Sparc, Grace Blackwell superchips)
- Evaluate, implement, and maintain state-of-the-art inference frameworks (TensorRT, Triton Inference Server, ONNX Runtime) to maximize GPU utilization and throughput
- Profile and optimize deep neural networks on NVIDIA GPUs using tools such as NVIDIA Nsight, PyTorch Profiler, and custom instrumentation
- Design and implement efficient model serving architectures that support both synchronous REST APIs and asynchronous batch processing workflows
- Collaborate with Platform and Edge Device teams to containerize inference systems (Docker, Kubernetes) for deployment across cloud and on-premise environments
- Partner with cloud providers (AWS, GCP, Azure) to optimize hosted inference solutions and leverage latest hardware accelerators
- Ensure inference systems meet regulatory requirements (FDA 510(k), SOC2) with comprehensive monitoring, logging, and audit capabilities
- Prototype and productionize new inference optimization techniques, including quantization, pruning, distillation, and dynamic batching strategies
- Build robust telemetry and monitoring systems to track model performance, latency, throughput, and resource utilization in production
Skills
- 7+ years of experience building and optimizing production ML inference systems at scale
- Expert-level proficiency in Python and experience writing high-performance inference services
- 5+ years of hands-on experience with PyTorch and at least one production inference tools (TensorRT, Triton Inference Server, ONNX Runtime, TorchServe)
- Deep understanding of computer vision model architectures, particularly generative models (GANs, diffusion models) and vision transformers
- Extensive experience profiling and optimizing deep neural networks on NVIDIA GPUs, including memory optimization, kernel fusion, and mixed-precision inference
- Strong background in image processing pipelines and libraries (OpenCV, Pillow, scikit-image) for handling large-scale medical imaging data
- Proven experience deploying ML systems on Kubernetes and major cloud providers (AWS, GCP, Azure)
- Experience with Docker containerization and orchestration for ML workloads
- Strong software engineering practices including version control (Git), CI/CD, unit testing, and production debugging
- Excellent communication, collaboration, and technical documentation skills
- Experience with medical imaging, digital pathology, or whole slide imaging (WSI) processing
- Knowledge of edge device deployment and embedded systems for AI inference
- Experience with MLOps tools (MLflow, Kubeflow, Apache Airflow) and model versioning
- Understanding of FDA regulatory requirements for AI/ML in medical devices
- Background in distributed inference systems and model parallelism techniques
- Familiarity with monitoring and logging tools (Prometheus, Grafana, ELK Stack)
Company Overview
Company H1B Sponsorship