---
title: "Anand (@anandms) | Hashnode"
description: "Software Engineer II @ Microsoft. Read the latest articles by Anand on Hashnode."
url: "https://hashnode.com/@anandms"
---

@anandms

# Anand

@anandms · Redmond, WA · Joined June 2023

Software Engineer II @ Microsoft

[Website](https://github.com/anandms101)

### [Someone Else's Words in Your Answer: A Design Review of Cloudflare's GLM-5.3 KV Cache Exposure](https://anandms101.hashnode.dev/someone-else-s-words-in-your-answer-a-design-review-of-cloudflare-s-glm-5-3-kv-cache-exposure?utm_source=hashnode&utm_medium=feed)

How a refused prompt crossed between two machines and came back with other customers' words, what Cloudflare and SGLang changed, and what's still open.

27 min read · [anandms101.hashnode.dev](https://anandms101.hashnode.dev/someone-else-s-words-in-your-answer-a-design-review-of-cloudflare-s-glm-5-3-kv-cache-exposure?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/someone-else-s-words-in-your-answer-a-design-review-of-cloudflare-s-glm-5-3-kv-cache-exposure/6aee9c260df03d4b598b1c33) [Nov 7](https://hashnode.com/posts/someone-else-s-words-in-your-answer-a-design-review-of-cloudflare-s-glm-5-3-kv-cache-exposure/6aee9c260df03d4b598b1c33)

### [Your AI Remembered. Did It Learn the Right Lesson?](https://anandms101.hashnode.dev/your-ai-remembered-did-it-learn-the-right-lesson?utm_source=hashnode&utm_medium=feed)

An agent can preserve your words but lose their limits. Following one travel request shows how records become lessons and reach the next task.

9 min read · [anandms101.hashnode.dev](https://anandms101.hashnode.dev/your-ai-remembered-did-it-learn-the-right-lesson?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/your-ai-remembered-did-it-learn-the-right-lesson/6adff09d0da1855b92970080) [Oct 27](https://hashnode.com/posts/your-ai-remembered-did-it-learn-the-right-lesson/6adff09d0da1855b92970080)

### [An Attention Window Is Not a Checkpoint: Inside vLLM's 128-Token Replay](https://anandms101.hashnode.dev/an-attention-window-is-not-a-checkpoint-inside-vllm-s-128-token-replay?utm_source=hashnode&utm_medium=feed)

A coding agent returns with another test result. Following that request reveals why vLLM saves less state between turns, then rebuilds a recent window without reproducing it exactly.

10 min read · [anandms101.hashnode.dev](https://anandms101.hashnode.dev/an-attention-window-is-not-a-checkpoint-inside-vllm-s-128-token-replay?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/an-attention-window-is-not-a-checkpoint-inside-vllm-s-128-token-replay/6acb2337061765bf8144b005) [Oct 11](https://hashnode.com/posts/an-attention-window-is-not-a-checkpoint-inside-vllm-s-128-token-replay/6acb2337061765bf8144b005)

### [The Model Is Untrusted Code: A Design Review of Meta's Muse](https://anandms101.hashnode.dev/the-model-is-untrusted-code-a-design-review-of-meta-s-muse?utm_source=hashnode&utm_medium=feed)

How Muse contains a prompt-injected agent, replayed against one malicious email, plus the questions I'd push on and the parts worth copying.

17 min read · [anandms101.hashnode.dev](https://anandms101.hashnode.dev/the-model-is-untrusted-code-a-design-review-of-meta-s-muse?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/the-model-is-untrusted-code-a-design-review-of-meta-s-muse/6abb3a812770f69963119bf4) [Sep 29](https://hashnode.com/posts/the-model-is-untrusted-code-a-design-review-of-meta-s-muse/6abb3a812770f69963119bf4)

### [Toon vs JSON: The Battle for Token Efficiency in LLM Applications](https://anandms101.hashnode.dev/toon-vs-json-the-battle-for-token-efficiency-in-llm-applications?utm_source=hashnode&utm_medium=feed)

Discover how Toon format reduces LLM token usage by 20% compared to JSON. Real data, cost analysis, and practical examples for OpenAI and Anthropic APIs. Sa

6 min read · [anandms101.hashnode.dev](https://anandms101.hashnode.dev/toon-vs-json-the-battle-for-token-efficiency-in-llm-applications?utm_source=hashnode&utm_medium=feed)

[Nov 8, 2025](https://hashnode.com/posts/toon-vs-json-the-battle-for-token-efficiency-in-llm-applications/690fc547dbea14e4c4fdf8ce)
