Medium

Syed Sadat Nazrul

Featured Book

Synapse: AI Crime Investigation

Synapse: AI Crime Investigation

An AI crime thriller about hackers, deepfakes, and a stolen biotech secret.

2024

Syed Sadat Nazrul

Syed Sadat Nazrul

Stories

Syed Sadat Nazrul

·

Nov 6

From Strawberry’s R’s to Hacking AI Agents

From Strawberry’s R’s to Hacking AI Agents

In

CodeToDeploy

by

Syed Sadat Nazrul

·

Nov 1

From Sand to Intelligence: The Global Supply Chain Behind AI

My usual blogs have been around AI algorithms with some sprinkle of philosophy. But given the current socail climate, I wanted add a bit…

From Sand to Intelligence: The Global Supply Chain Behind AI

Syed Sadat Nazrul

·

Oct 6

Technology, Progress, and the Problem of Human Values

Needless to say, technological progress over the past two years has been both exciting and overwhelming. For starters, as an individual…

Technology, Progress, and the Problem of Human Values

Syed Sadat Nazrul

·

Sep 20

Your AI Didn’t “Escape.” Someone Let It Out.

AI escape stories obscure the real questions: objectives, permissions, failed safeguards, and who is accountable when systems cause harm.

Your AI Didn’t “Escape.” Someone Let It Out.

Syed Sadat Nazrul

·

Jun 21

History Repeats: Vibe Coding Still Needs Unix Philosophy

Almost every argument I have heard against vibe coding, I have heard before. Too verbose. Too messy. Not real programming, since you are…

History Repeats: Vibe Coding Still Needs Unix Philosophy

Syed Sadat Nazrul

·

May 21

From Popularity to Personalization: Target Design in Search, Recommendations, and RAG

Every time you buy something on Amazon, watch something on Netflix, click a search result on Google, or get an answer from ChatGPT, a…

From Popularity to Personalization: Target Design in Search, Recommendations, and RAG

Syed Sadat Nazrul

·

May 20

Turning LLM Outputs Into Production Systems

What it actually takes to move from an impressive LLM demo to a production system teams can trust.

Turning LLM Outputs Into Production Systems

Syed Sadat Nazrul

·

May 20

You Shipped an LLM Feature. Now How Do You Know It Is Working?

Shipping an LLM feature feels like the finish line. You got stakeholder buy-in, you built the pipeline, you ran the tests, you deployed it…

You Shipped an LLM Feature. Now How Do You Know It Is Working?

Syed Sadat Nazrul

·

May 19

The Hardest Part of Cybersecurity ML Is Not the Model. It Is the Target.

Building cybersecurity ML systems under real-world constraints, noisy labels, and incomplete observability.

The Hardest Part of Cybersecurity ML Is Not the Model. It Is the Target.

Syed Sadat Nazrul

·

May 18

Thinking Fast and Slow in Cybersecurity ML: The Case for Interpretability

Thinking Fast and Slow in Cybersecurity ML: The Case for Interpretability

Syed Sadat Nazrul

Syed Sadat Nazrul

Book Author

10 years shipping ML at scale: HPC, cybersecurity, fraud detection, marketplace discovery. Writing about production AI strategy. Opinions are my own.

Following

Help

Status

About

Careers

Press

Blog

Store

Privacy

Rules

Terms

Text to speech