Netflix
Senior
Build the infrastructure that trains, aligns, and serves Netflix's most critical ML models
Netflix's Model Runtime team owns the systems behind training, alignment, and serving of Netflix's core ML models, spanning RL-based post-training, distributed training at scale, and next-gen GenAI inference. This role sits at the intersection of systems engineering and ML, working across the stack from PyTorch internals down to GPU kernels.
Step into this interview
Free · a live voice mock calibrated to this exact role
What this interview tests
- Distributed training systems (FSDP, tensor/pipeline/context parallelism)
- RL-based post-training infra (GRPO, DPO, PPO, reward modeling)
- GPU performance profiling and optimization (CUDA, NCCL, Nsight, PyTorch profiler)
- Inference optimization (vLLM, TensorRT, quantization, KV-cache, continuous batching)
- Multimodal/diffusion model serving and generation pipelines
- Operational excellence: observability, logging, on-call for ML infra
Common question themes
Design a fault-tolerant distributed training system across hundreds of GPUs
How would you build RL/DPO-style post-training infra for a recommendation model
Describe a time you profiled and optimized a training or inference workload from framework down to kernel level
How do you evaluate new hardware/accelerators or frameworks for an ML infra stack
Trade-offs between real-time, near-real-time, and batch inference for GenAI workloads
How do you operate a small, highly autonomous infra team with outsized impact
Related interviews
Netflix
Senior
Integrations Support Engineer 5 - Ads Conversion API
Netflix
Senior
Software Engineering 5 - Ads Conversion Attribution
Netflix
Senior
Software Engineer 5, Ads Reporting
Cohere
Staff
Software Engineer, GPU Infrastructure (HPC)
Amazon
Mid
Software Development Engineer I, ML Infra Services, Annapurna Labs
Amazon
New grad