---
title: "Harshvardhan Singh (@hrsvd) | Hashnode"
description: "Writing what matters. Read the latest articles by Harshvardhan Singh on Hashnode."
url: "https://hashnode.com/@hrsvd"
---

@hrsvd

# Harshvardhan Singh

@hrsvd · Joined October 2025

Writing what matters

### [10 Production Problems That Decide Whether Your RAG or Agent System Survives Contact With Real Users](https://hrsvd.hashnode.dev/10-production-problems-in-rag?utm_source=hashnode&utm_medium=feed)

Detailed solution of problems faced in RAG and Agentic AI in production.

44 min read · [hrsvd.hashnode.dev](https://hrsvd.hashnode.dev/10-production-problems-in-rag?utm_source=hashnode&utm_medium=feed)

[Sep 12](https://hashnode.com/posts/10-production-problems-in-rag/6aa5061b3618fb2640ebf8b0)

### [Semantic Caching in Production: What Breaks When Cosine Similarity Isn't Enough](https://hrsvd.hashnode.dev/semantic-caching-in-production-what-breaks-when-cosine-similarity-isn-t-enough?utm_source=hashnode&utm_medium=feed)

The 3 AM Page

7 min read · [hrsvd.hashnode.dev](https://hrsvd.hashnode.dev/semantic-caching-in-production-what-breaks-when-cosine-similarity-isn-t-enough?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/semantic-caching-in-production-what-breaks-when-cosine-similarity-isn-t-enough/6a70534568d5fc66a416ca64) [Aug 3](https://hashnode.com/posts/semantic-caching-in-production-what-breaks-when-cosine-similarity-isn-t-enough/6a70534568d5fc66a416ca64)

### [Why Java Is Quietly Becoming a Serious Language for AI Infrastructure](https://hrsvd.hashnode.dev/why-java-is-quietly-becoming-a-serious-language-for-ai-infrastructure?utm_source=hashnode&utm_medium=feed)

Discover why Java is becoming a serious choice for AI infrastructure, RAG systems, LLM applications, and enterprise AI backends.

9 min read · [hrsvd.hashnode.dev](https://hrsvd.hashnode.dev/why-java-is-quietly-becoming-a-serious-language-for-ai-infrastructure?utm_source=hashnode&utm_medium=feed)

[Jul 20](https://hashnode.com/posts/why-java-is-quietly-becoming-a-serious-language-for-ai-infrastructure/6a5de4cc5cbc31410371acc8)

### [Context Engineering: A Reference Architecture for Production LLM Systems](https://hrsvd.hashnode.dev/context-engineering-a-reference-architecture-for-production-llm-systems?utm_source=hashnode&utm_medium=feed)

Why the unit of engineering effort has moved from the instruction to the pipeline, and what a mature context architecture actually looks like

14 min read · [hrsvd.hashnode.dev](https://hrsvd.hashnode.dev/context-engineering-a-reference-architecture-for-production-llm-systems?utm_source=hashnode&utm_medium=feed)

[Jul 11](https://hashnode.com/posts/context-engineering-a-reference-architecture-for-production-llm-systems/6a522c5055a9008ef4b275eb)

### [Why LLMs Forget Everything: A Developer's Guide to Context Windows, RAG, and Memory Architectures](https://hrsvd.hashnode.dev/why-llms-forget-everything-a-developer-s-guide-to-context-windows-rag-and-memory-architectures?utm_source=hashnode&utm_medium=feed)

A practical breakdown of why "stateless by default" is the defining constraint of LLM applications and the architectural patterns used to work around it.

9 min read · [hrsvd.hashnode.dev](https://hrsvd.hashnode.dev/why-llms-forget-everything-a-developer-s-guide-to-context-windows-rag-and-memory-architectures?utm_source=hashnode&utm_medium=feed)

[2](https://hashnode.com/posts/why-llms-forget-everything-a-developer-s-guide-to-context-windows-rag-and-memory-architectures/6a4e333ac3e9796d5f573ab0) [Jul 8](https://hashnode.com/posts/why-llms-forget-everything-a-developer-s-guide-to-context-windows-rag-and-memory-architectures/6a4e333ac3e9796d5f573ab0)