Netflix Open-Sources Internal LLM Inference Stack
AxSaucedo · x · 2026-07-20
Netflix shared an overview of how they support various internal use cases with a large-scale LLM inference stack. The accompanying diagram illustrates the overall service architecture: consumer applications first route through a unified service for preprocessing and post-processing; inference is then handled by the NVIDIA Triton Inference Server paired with vLLM, while also accessing multiple data sources for feature and data retrieval.
This diagram depicts a classic enterprise-grade unified serving system: routing traffic by use case and A/B testing, decoupling pre/post-processing from inference, and isolating the model scoring service.
Related event: Netflix Reveals Internal LLM Inference Stack(2 posts)→
More from Infra
- Anthropic S-1 reveals $800B+ in off-books future compute commitments — GaryMarcus · 2026-10-27
- ~42% of Anthropic's ~11GW contracted compute is NVIDIA GPUs — ns123abc · 2026-10-27
- EdgeRock-4B holds KV cache at 72 MiB across 256K context, 99.1% smaller than Qwen3.5-4B — alifcoder · 2026-10-27
- RTX 5080 price jumps from $1658 to $2318 in ten days on Amazon — Frizzy-MacDrizzle · 2026-10-27
- ComfyUI on 12GB VRAM: measured peaks show offloading turns 24s into 11 minutes — Impossible-Mail-4619 · 2026-10-27
- Neon rebuilds its docs search entirely on its own stack: Lakebase Search, Functions, AI Gateway — matei_zaharia · 2026-10-27