---
title: "himanshu (@hmnshpathak1) | Hashnode"
description: "Building scalable systems, one architecture decision at a time. Read the latest articles by himanshu on Hashnode."
url: "https://hashnode.com/@hmnshpathak1"
---

@hmnshpathak1

# himanshu

@hmnshpathak1 · Delhi, India · Joined June 2026

Building scalable systems, one architecture decision at a time.

### [How to Choose the Right Search Model for RAG Using Retrieval Metrics](https://himanshutech.hashnode.dev/how-to-choose-the-right-search-model-for-rag-using-retrieval-metrics?utm_source=hashnode&utm_medium=feed)

Use evaluation — not hype — to match your embedding or search approach to how users actually query

10 min read · [himanshutech.hashnode.dev](https://himanshutech.hashnode.dev/how-to-choose-the-right-search-model-for-rag-using-retrieval-metrics?utm_source=hashnode&utm_medium=feed)

[Jul 3](https://hashnode.com/posts/how-to-choose-the-right-search-model-for-rag-using-retrieval-metrics/6a47c8a080cf8f787fe18603)

### [The hidden cost of RAG: our vector database is not the bottleneck](https://himanshutech.hashnode.dev/the-hidden-cost-of-rag-our-vector-database-is-not-the-bottleneck?utm_source=hashnode&utm_medium=feed)

In AI product search , vector databases rarely bottleneck RAG systems. Learn where retrieval-augment generation actually gets slow and expensive

10 min read · [himanshutech.hashnode.dev](https://himanshutech.hashnode.dev/the-hidden-cost-of-rag-our-vector-database-is-not-the-bottleneck?utm_source=hashnode&utm_medium=feed)

[Jun 27](https://hashnode.com/posts/the-hidden-cost-of-rag-our-vector-database-is-not-the-bottleneck/6a3ff083e0ef93631af351ee)

### [What AI Engineers Can Learn From OpenSearch's Move to gRPC](https://himanshutech.hashnode.dev/what-ai-engineers-can-learn-from-opensearch-s-move-to-grpc?utm_source=hashnode&utm_medium=feed)

OpenSearch is adopting gRPC to improve performance and scalability. Here's what AI engineers can learn about low-latency communication, distributed systems, and building AI infrastructure at scale.

9 min read · [himanshutech.hashnode.dev](https://himanshutech.hashnode.dev/what-ai-engineers-can-learn-from-opensearch-s-move-to-grpc?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/what-ai-engineers-can-learn-from-opensearch-s-move-to-grpc/6a368e34998af5167ddb35b9) [Jun 20](https://hashnode.com/posts/what-ai-engineers-can-learn-from-opensearch-s-move-to-grpc/6a368e34998af5167ddb35b9)

### [Lambda vs. EKS for AI Orchestration at Scale](https://himanshutech.hashnode.dev/lambda-vs-eks-for-ai-orchestration-at-scale?utm_source=hashnode&utm_medium=feed)

Building AI at scale? Compare AWS Lambda and Amazon EKS across scalability, latency, operational complexity, and cost with real-world insights.

8 min read · [himanshutech.hashnode.dev](https://himanshutech.hashnode.dev/lambda-vs-eks-for-ai-orchestration-at-scale?utm_source=hashnode&utm_medium=feed)

[Jun 13](https://hashnode.com/posts/lambda-vs-eks-for-ai-orchestration-at-scale/6a2d99b2a26f215c3a089015)