---
title: "Shubham Tatvamasi (@ShubhamTatvamasi) | Hashnode"
description: "AI infrastructure engineer. Inference, GPUs and GitOps on Kubernetes."
url: "https://hashnode.com/@ShubhamTatvamasi"
---

@ShubhamTatvamasi

# Shubham Tatvamasi

@ShubhamTatvamasi · Joined January 2023

[Website](https://shubhamtatvamasi.com/)

### [Inside NVIDIA Dynamo: A Deep Technical Tour of the Datacenter-Scale Inference Stack](https://blog.shubhamtatvamasi.com/inside-nvidia-dynamo-deep-dive?utm_source=hashnode&utm_medium=feed)

Disaggregated prefill/decode, KV-aware routing, NIXL, KVBM, the SLA Planner, and the Kubernetes machinery behind them

29 min read · [blog.shubhamtatvamasi.com](https://blog.shubhamtatvamasi.com/inside-nvidia-dynamo-deep-dive?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/inside-nvidia-dynamo-deep-dive/6afc4f47fb30ba10dd112f43) [4d ago](https://hashnode.com/posts/inside-nvidia-dynamo-deep-dive/6afc4f47fb30ba10dd112f43)

### [vLLM Deep Dive: How PagedAttention, Continuous Batching and the V1 Engine Serve LLMs in Production](https://blog.shubhamtatvamasi.com/vllm-deep-dive-how-pagedattention-continuous-batching-and-the-v1-engine-serve-llms-in-production?utm_source=hashnode&utm_medium=feed)

Every layer above it routes, scales and governs requests. vLLM is the layer that turns GPU memory into tokens.

52 min read · [blog.shubhamtatvamasi.com](https://blog.shubhamtatvamasi.com/vllm-deep-dive-how-pagedattention-continuous-batching-and-the-v1-engine-serve-llms-in-production?utm_source=hashnode&utm_medium=feed)

[3](https://hashnode.com/posts/vllm-deep-dive-how-pagedattention-continuous-batching-and-the-v1-engine-serve-llms-in-production/6af1a0c4a32bf9f449f76bd3) [Nov 9](https://hashnode.com/posts/vllm-deep-dive-how-pagedattention-continuous-batching-and-the-v1-engine-serve-llms-in-production/6af1a0c4a32bf9f449f76bd3)

### [KAI Scheduler Deep Dive: How NVIDIA Schedules GPUs on Kubernetes](https://blog.shubhamtatvamasi.com/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes?utm_source=hashnode&utm_medium=feed)

Queues, gang scheduling, fractional GPUs and topology-aware placement: how NVIDIA’s open-source scheduler shares GPUs

35 min read · [blog.shubhamtatvamasi.com](https://blog.shubhamtatvamasi.com/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes?utm_source=hashnode&utm_medium=feed)

[3](https://hashnode.com/posts/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes/6ad9850f035d0b8ec4de8b8f) [Oct 22](https://hashnode.com/posts/kai-scheduler-deep-dive-how-nvidia-schedules-gpus-on-kubernetes/6ad9850f035d0b8ec4de8b8f)

### [NVIDIA Dynamo vs KServe: Which Layer Should Own Your LLM Inference?](https://blog.shubhamtatvamasi.com/nvidia-dynamo-vs-kserve-which-layer-should-own-your-llm-inference?utm_source=hashnode&utm_medium=feed)

Cache-aware routing, prefill/decode disaggregation and autoscaling compared, and how to choose per model pool

13 min read · [blog.shubhamtatvamasi.com](https://blog.shubhamtatvamasi.com/nvidia-dynamo-vs-kserve-which-layer-should-own-your-llm-inference?utm_source=hashnode&utm_medium=feed)

[Oct 19](https://hashnode.com/posts/nvidia-dynamo-vs-kserve-which-layer-should-own-your-llm-inference/6ad6439fa7db020af3c69de6)

### [Running LLMs on Kubernetes in Production: KServe, vLLM, llm-d, Envoy AI Gateway, and LiteLLM](https://blog.shubhamtatvamasi.com/running-llms-on-kubernetes-in-production-kserve-vllm-llm-d-envoy-ai-gateway-and-litellm?utm_source=hashnode&utm_medium=feed)

One OpenAI-compatible endpoint, model-aware routing, GPU-aware scheduling, and scaling driven by inference pressure, not CPU

26 min read · [blog.shubhamtatvamasi.com](https://blog.shubhamtatvamasi.com/running-llms-on-kubernetes-in-production-kserve-vllm-llm-d-envoy-ai-gateway-and-litellm?utm_source=hashnode&utm_medium=feed)

[Oct 11](https://hashnode.com/posts/running-llms-on-kubernetes-in-production-kserve-vllm-llm-d-envoy-ai-gateway-and-litellm/6acb0e343177a7b8a6788104)
