MLOps Engineer

Hi!

Thank you for taking some time to look at our requisition. We are a US-based company working on an AI product in the entertainment space. Our app is geared towards children and we are working with major film companies.

We have a strong, distributed team, mostly in Europe. We're looking for an experienced person to help use with MLOps.

Core MLOps Responsibilities:

• Model Deployment: Convert ComfyUI workflows to production Python pipelines
• Infrastructure Management: Multi-provider GPU orchestration (RunPod + future providers)
• CI/CD for ML: Automated model deployment and rollback systems
• Monitoring & Observability: Pipeline performance, model drift, and system health
• Scalability: Serverless GPU management and load balancing
• Model Lifecycle: Version control, and hot-swapping of LoRAs

AI/ML Pipeline (Critical):

• Deep experience with Diffusion models (Stable Diffusion, Flux)
• Hands-on ComfyUI to Python conversion experience
• Computer vision libraries: OpenCV, PIL, torchvision
• Model inference optimization (batching, memory management)
• Experience with diffuser library
• Experience with ControlNets, LoRA, and inpainting workflows
• Experience with GroundingDINO, SAM

Backend Development:

• FastAPI/Python (mid/senior level)
• Async programming and queue management
• PostgreSQL/AlloyDb
• RESTful API design with proper error handling

DevOps/Infrastructure:

• Docker containerization
• Google Cloud Platform (GCS, Cloud Run, CloudBuild)
• Git Actions
• CI/CD pipeline setup
• GPU Providers Platform (RunPod nice to have)

GPU/Serverless:

• RunPod API integration (preferred) or other GPU providers
• GPU memory optimization
• Cold start minimization strategies
• Multi-provider orchestration patterns

Monitoring & Observability:

• Custom metrics for ML pipelines
• Performance monitoring and alerting
• Integration with data warehouse systems

Nice-to-Have:

• Previous work with content generation platforms
• Experience with model serving frameworks (TorchServe, TensorRT)
• Experience with training/fine-tuning image generation models (e.g, Stable Diffusion, Flux with LoRA)