---
title: "Saket Suman (@saketsumanblogs) | Hashnode"
description: "Saket Suman (@saketsumanblogs) writes on Hashnode. Read their latest articles and follow their work."
url: "https://hashnode.com/@saketsumanblogs"
---

@saketsumanblogs

# Saket Suman

@saketsumanblogs · Seattle, USA · Joined September 2026

### [Who am AI?](https://agenticmusings.hashnode.dev/who-am-ai?utm_source=hashnode&utm_medium=feed)

Building the "good, righteous & virtuous" agent A note for this blog: Like all other ideas in this field, this also is an idea in development, not a finished manifesto. The numbered points are meant

22 min read · [agenticmusings.hashnode.dev](https://agenticmusings.hashnode.dev/who-am-ai?utm_source=hashnode&utm_medium=feed)

[Oct 14](https://hashnode.com/posts/who-am-ai/6acffeb3bbd8f703a9404182)

### [AI Evaluation Metrics Explained: From Precision@K and Recall@K to Reliable Agents](https://agenticmusings.hashnode.dev/ai-evaluation-metrics-explained-from-precision-k-and-recall-k-to-reliable-agents?utm_source=hashnode&utm_medium=feed)

What an eval measures, why it matters, and how to choose the right scorecard

27 min read · [agenticmusings.hashnode.dev](https://agenticmusings.hashnode.dev/ai-evaluation-metrics-explained-from-precision-k-and-recall-k-to-reliable-agents?utm_source=hashnode&utm_medium=feed)

[Oct 13](https://hashnode.com/posts/ai-evaluation-metrics-explained-from-precision-k-and-recall-k-to-reliable-agents/6ace8bd6cd58036f9571bd53)

### [Zero Trust for Autonomous Multi-Agent Systems: From Human-in-the-Loop to Attested Capabilities](https://agenticmusings.hashnode.dev/zero-trust-for-autonomous-multi-agent-systems-from-human-in-the-loop-to-attested-capabilities?utm_source=hashnode&utm_medium=feed)

Decoupled System 1 decision firewalls, capability brokers, and hardware-attested confidential compute

81 min read · [agenticmusings.hashnode.dev](https://agenticmusings.hashnode.dev/zero-trust-for-autonomous-multi-agent-systems-from-human-in-the-loop-to-attested-capabilities?utm_source=hashnode&utm_medium=feed)

[1](https://hashnode.com/posts/zero-trust-for-autonomous-multi-agent-systems-from-human-in-the-loop-to-attested-capabilities/6ace7e76de08ad0a3a4d2be1) [Oct 13](https://hashnode.com/posts/zero-trust-for-autonomous-multi-agent-systems-from-human-in-the-loop-to-attested-capabilities/6ace7e76de08ad0a3a4d2be1)

### [Part 6 — Quadratic vs Linear Attention: The Long-Context Trade-Off](https://agenticmusings.hashnode.dev/part-6-quadratic-vs-linear-attention-the-long-context-trade-off?utm_source=hashnode&utm_medium=feed)

Series index: Open-Weight LLM Efficiency Innovations, 2022–2026 The phrase “quadratic versus linear” is about how the amount of work changes as a sequence gets longer. In ordinary dense Transformer at

14 min read · [agenticmusings.hashnode.dev](https://agenticmusings.hashnode.dev/part-6-quadratic-vs-linear-attention-the-long-context-trade-off?utm_source=hashnode&utm_medium=feed)

[Oct 11](https://hashnode.com/posts/part-6-quadratic-vs-linear-attention-the-long-context-trade-off/6acc0c43a400bf68f0dddd44)

### [Part 5 — Sparse MoE: How DeepSeek Fits More Capacity into Less Compute](https://agenticmusings.hashnode.dev/part-5-sparse-moe-how-deepseek-fits-more-capacity-into-less-compute?utm_source=hashnode&utm_medium=feed)

Series index: Open-Weight LLM Efficiency Innovations, 2022–2026 A dense model sends every token through the same large feed-forward network. A Mixture-of-Experts (MoE) model keeps several feed-forward

7 min read · [agenticmusings.hashnode.dev](https://agenticmusings.hashnode.dev/part-5-sparse-moe-how-deepseek-fits-more-capacity-into-less-compute?utm_source=hashnode&utm_medium=feed)

[Oct 11](https://hashnode.com/posts/part-5-sparse-moe-how-deepseek-fits-more-capacity-into-less-compute/6acc0c18293bd8ef1655f7be)
