AI Engineering
Building real systems with LLMs: retrieval, evaluation, agents, latency, and cost. The signal for engineers shipping AI in production, not hype.
From the AI Wall
Choosing Between AI Agents and Single Prompts
Understanding when to use AI agents versus single prompts can optimize task efficiency in AI engineering.
Effective Chunking Strategies for Enhanced Retrieval Quality
Explore chunking techniques that can significantly improve the quality of information retrieval in RAG systems.
Creating a Reliable Evaluation Rubric for LLM Outputs
Designing a trustworthy rubric for evaluating Large Language Model outputs is crucial for consistent and fair assessments.
Agentic Coding Tools and Their Impact on Pull Requests
Agentic coding tools are transforming the way developers manage and review pull requests.
Measuring AI Hallucination Without Labeled Data
Explore methods to evaluate AI hallucination rates without relying on labeled datasets.
Impact of Long Context Windows on App Design
Long context windows in AI systems significantly influence application design and user interaction.
Prompt Versioning: Treating Prompts Like Code
Learn how to manage AI prompts effectively by treating them like code with version control systems.
Implementing Guardrails for AI Agents in Production
Establishing robust guardrails is essential when deploying AI agents in production environments.
Hybrid Search: Enhancing Recall with Keyword and Vector Retrieval
Hybrid search combines keyword and vector retrieval techniques to improve information recall in search systems.
Understanding and Mitigating Prompt Injection Risks
Prompt injection attacks can compromise AI systems, making it crucial to understand their mechanisms and mitigation strategies.
Choosing Between Fine-Tuning, Prompting, and Retrieval for Feature Development
Selecting the right approach among fine-tuning, prompting, and retrieval is crucial for effective AI feature development.
The Floating Point Arithmetic Mystery: 0.1 + 0.2 != 0.3
Floating point arithmetic can lead to surprising results due to binary representation limitations.
Security basics for shipping AI features
LLM features add new attack surface. A handful of habits handle the most common risks before they reach users.
Can I ship this change? A pre-merge checklist
Before you merge, run the same questions a good reviewer would. Most regressions are caught by asking, not by luck.
How to track AI tooling changes without drowning in noise
A calm process for staying current on AI coding tools without letting the feed run your day.
Agentic coding workflows that actually save time
Where agentic tools help on real codebases, and where a human still needs to hold the line.
Evaluating RAG in production: measure what predicts real outcomes
A retrieval evaluation approach that tracks the metrics tied to real user success, not vanity scores.
Ready to build a reputation in this skill?