https://blog.mikegchambers.com/Personal notes and observations on artificial intelligence, machine learning, and building with foundation models. 2026-03-20T06:16:28+00:00 Mike Chambers https://blog.mikegchambers.com/ Jekyll © 2026 Mike Chambers /assets/img/favicons/favicon.ico /assets/img/favicons/favicon-96x96.png 2026-03-20T00:00:00+00:00 2026-03-20T00:00:00+00:00 https://blog.mikegchambers.com/posts/llm-em-dash-obsession/ Mike Chambers

I tested 27 models on Amazon Bedrock for em dash usage. Llama produces zero. Claude and Palmyra can't stop. What does that tell us about how LLMs learn style?

2026-03-19T00:00:00+00:00 2026-03-19T00:00:00+00:00 https://blog.mikegchambers.com/posts/mcp-tool-protection/ Mike Chambers

My AI agent has access to my email and Slack. Here are four tactics I use to stop it from sending a career-ending message — from system prompts to deterministic hooks, LLM-as-a-judge steering, and Cedar policies at cloud scale.

2026-03-18T00:00:00+00:00 2026-03-18T03:29:57+00:00 https://blog.mikegchambers.com/posts/production-browser-agent-aws/ Mike Chambers

How to wire together AgentCore Browser, Nova Act, and IAM authentication to build a production browser agent on AWS, with complete working code.

2026-03-12T00:00:00+00:00 2026-03-18T03:29:57+00:00 https://blog.mikegchambers.com/posts/pirate-lora-fine-tuning/ Mike Chambers

I fine-tuned Qwen2.5-0.5B to always talk like a pirate using LoRA. The first attempt failed because of a system prompt in the training data. Here's what I learned about training data design.

2026-03-11T00:00:00+00:00 2026-03-18T03:29:57+00:00 https://blog.mikegchambers.com/posts/agentic-framework-landscape/ Mike Chambers

Nine Python agent frameworks, compared honestly. Architecture, code samples, community sentiment, and what actually matters when you're picking one.