---
title: "Sagnik Das (@sagnikdas) | Hashnode"
description: "Explaining modern AI systems, open models, and how they actually work. Read the latest articles by Sagnik Das on Hashnode."
url: "https://hashnode.com/@sagnikdas"
---

@sagnikdas

# Sagnik Das

@sagnikdas · Nürnberg, BY, Deutschland · Joined July 2025

Explaining modern AI systems, open models, and how they actually work.

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

### [Inside GPT-OSS: Open-Weight Reasoning Models Built for Agentic AI](https://blogs.sagnikdas.com/inside-gpt-oss-open-weight-reasoning-models-built-for-agentic-ai?utm_source=hashnode&utm_medium=feed)

GPT-OSS-20B and GPT-OSS-120B

8 min read · [blogs.sagnikdas.com](https://blogs.sagnikdas.com/inside-gpt-oss-open-weight-reasoning-models-built-for-agentic-ai?utm_source=hashnode&utm_medium=feed)

[Jan 22](https://hashnode.com/posts/inside-gpt-oss-open-weight-reasoning-models-built-for-agentic-ai/6971cb02562037dd4faa4c16)

### [Why Variational Autoencoders Are the Secret Sauce of the GenAI Revolution](https://blogs.sagnikdas.com/variational-autoencoders-secret-sauce-genai?utm_source=hashnode&utm_medium=feed)

Unlocking Generative AI

8 min read · [blogs.sagnikdas.com](https://blogs.sagnikdas.com/variational-autoencoders-secret-sauce-genai?utm_source=hashnode&utm_medium=feed)

[Jul 31, 2025](https://hashnode.com/posts/variational-autoencoders-secret-sauce-genai/688b4164c0df5a60408f6f8f)

### [Your 92% Accurate AI Model Might Be Dangerous (Here's Why)](https://blogs.sagnikdas.com/your-92-accurate-ai-model-might-be-dangerous-heres-why?utm_source=hashnode&utm_medium=feed)

Originally an epistemology paper, adapted for the blog community

3 min read · [blogs.sagnikdas.com](https://blogs.sagnikdas.com/your-92-accurate-ai-model-might-be-dangerous-heres-why?utm_source=hashnode&utm_medium=feed)

[Jul 16, 2025](https://hashnode.com/posts/your-92-accurate-ai-model-might-be-dangerous-heres-why/6887c2ca3617f09e31b2a6eb)
