Bin Zhang Data, AI, and engineering notes

Bin Zhang's Field Notes

Modern notes for data, AI, and engineering systems.

Practical essays, implementation notes, and mental models for LLMs, infrastructure, data platforms, and the craft of building durable software.

Browse posts View archive

Pinned dispatch

Pinned

Mar 7, 2026

From Prompt to Response: A Step-by-Step Walkthrough of LLM Inference

LARGE LANGUAGE MODELS

Update: For a deeper systems-level treatment of LLM inference, especially the interaction between request scheduling, prefill, decode, and KV-cache reuse, see arXiv:2606.24937.

Read article

Fresh from the notebook

Recent writing

All posts

Jun 23, 2026

Use Databricks Models with VS Code Copilot and Copilot CLI

AI ENGINEERINGDATA ENGINEERING

I wanted one Databricks-hosted model to work in two developer surfaces:

Read note

Jun 18, 2026

A Git-Native Message Channel for Local Coding Agents

AI ENGINEERINGSOFTWARE ENGINEERING

My previous local development workflow was simple:

Read note

Mar 15, 2026

Attention Dilution

LARGE LANGUAGE MODELS

Attention dilution (also called context dilution) is one of the fundamental limitations of transformer-based LLMs when dealing with long...

Read note

Mar 8, 2026

AI Terminology: Agents, Skills, RAG, MCP, and the Layers Beneath the Hype

AI ENGINEERING

How many of these terms do you actually recognize?

Read note

Jan 4, 2026

ChatGPT in 2025: A Year in Review

LARGE LANGUAGE MODELS

ChatGPT Stats ChatGPT Growth ChatGPT Revenue

Read note

Nov 27, 2025

The Mandate for Leadership in AI Engineering

AI ENGINEERING

Over the next 12 to 24 months, the differentiator among engineers will shift from mastery of programming languages like Rust, Go, or Python,...

Read note