Writing about development and technology.
Shipping an AI feature and calling it good because it 'seems to work' is not a quality strategy. This post covers types of LLM evals, key metrics, building evaluation datasets, eval tools like promptfoo and Braintrust, and integrating evals into CI.
Beyond basic prompting — how to get reliably structured output from LLMs. System/user/assistant role design, few-shot examples, chain-of-thought, JSON mode, function calling, and building type-safe LLM responses with Zod and the AI SDK.
LLMs don't know what they weren't trained on. Here's how RAG fixes that — walking through the complete pipeline from document ingestion to chunking, embedding, vector storage, retrieval, and generation with real Python and TypeScript examples.
Both are children of Transformer, so why the difference? Using 'Fill-in-the-blank' vs 'Write-next-word' analogies to explain BERT vs GPT. Practical guide based on trial and error.
Overcoming RNN's 'dementia' and how Google flipped the world with 'Attention Is All You Need'. From Query-Key-Value library analogies to Multi-Head Attention and Vision Transformers.
A practical guide to choosing between Supervised, Unsupervised, and Semi-Supervised Learning when you don't have labeled data.
Understanding AI, Machine Learning, Deep Learning, and Generative AI. Deep dive into Transformer architecture, RAG vs Fine-tuning, Ethical AI, and a practical roadmap for developers transitioning to AI Engineering.
