How I Fine-Tuned TinyLlama on My Laptop Using LoRA
A complete walkthrough of fine-tuning a 1.1B parameter model on a Mac using LoRA, merging the adapter, and serving it locally with Ollama.
ReadBuilding the Harness: How Real ML Systems Actually Run Models
You trained the model. Now what? This post explains model inference and harness engineering with real code and zero hand-waving.
ReadCognitive Offloading vs Cognitive Surrender: One Makes Sense, the Other Doesn't
Cognitive offloading is smart. Cognitive surrender is dangerous. A Wharton study of 1,372 people shows most of us have stopped telling the difference.
ReadThe Person Who Writes the Best Skills Wins
Skills are becoming the new language for AI tools. Write a better skill and the LLM doesn't just follow instructions, it starts thinking like you.
ReadAI Tells You What to Build. Engineers Know Why It's Complicated.
AI can generate a product plan in minutes. That doesn't mean it can be shipped in the same time. Here's the part the hype doesn't talk about.
ReadHow I used Logistic Regression to Calculate Compatibility Scores in a Dating App
How can a simple statistical model make digital interactions feel personal? In a dating app project I worked on, I used logistic regression to turn user answers into meaningful insights. This blog shares my approach and the lessons I learned along the way.
Read