@mdazlaanzubair

Muhammad Azlaan Zubair

@mdazlaanzubairKarachi, PakistanJoined December 2022

Software Engineer | Applied AI Researcher | AI Workflows & Automation

About

Engineer. Researcher. Builder. I build things for web, experiment with AI, and interested in research.

Available for

Open to building products, collaborating on research, and sharing ideas.

Muhammad Azlaan Zubair's blogs

Wait, What?!blog.mdazlaanzubair.com22 posts

Articles Comments1

Recently published

MAMuhammad Azlaan Zubairblog.mdazlaanzubair.com3d ago · 10 min read

GPT-6 Astra Is More Aligned. I Still Wouldn’t Trust It With More Autonomy.

I would let GPT-6 Astra do more work. I would not let GPT-6 Astra decide what it is allowed to do. That distinction is more interesting to me than most of the benchmark numbers in OpenAI's launch. Op

MAMuhammad Azlaan Zubairblog.mdazlaanzubair.com5d ago · 6 min read

The Model Is Not the Architecture

The most visible AI product is a chat box. The most instructive AI systems often have no chat box at all. Look across Google DeepMind's work on algorithms, quantum error correction, robotics, and hist

MAMuhammad Azlaan Zubairblog.mdazlaanzubair.comSep 1 · 6 min read

An Agent API Migration Is an Ownership Audit

An API migration can compile, return the expected text, and still be wrong. That is the real risk in moving from OpenAI's Assistants API to Responses. The old objects do not map to new objects without

MAMuhammad Azlaan Zubairblog.mdazlaanzubair.comAug 30 · 7 min read

A Tool Result Is Evidence, Not Ground Truth

Connecting an agent to a database does not remove hallucination risk. It gives the system a second source that can be stale, malformed, unauthorized, or simply wrong—and a protocol that may make the m

MAMuhammad Azlaan Zubairblog.mdazlaanzubair.comAug 27 · 6 min read

Code Generation Is Fast. Verification Is Not.

Agentic IDEs can produce a runnable full-stack application in minutes. That does not mean the application became cheap to trust. In a controlled study of 48 generated projects, the researchers spent a