Writing
18 posts
on AI, ML and building
Long-form and code-first: LLMs, agents, evals, context and the bugs that only show up with real users. Every number in these posts came from code that actually ran.
2026
6 posts
- BuildingWhat shipping four products after hours taught me
- AI AgentsA WhatsApp AI booking agent in Python that can't double-book
- AI EngineeringAI agents vs workflows: when to use an agent, and when not
- React NativeLarge file uploads in React Native: two iOS bugs, 377 videos
- EngineeringSafe to spend: the maths behind one daily budget number
- machine learningData leakage in machine learning: 5 examples in scikit-learn
2025
6 posts
- LLMsRun small LLMs locally with Ollama: when 3B is enough
- LLMsHow to evaluate LLM outputs in Python: a starter kit
- AI EngineeringWhat is context engineering? A practical guide with code
- AI EngineeringHow to use AI coding agents effectively
- MCPHow to build an MCP server in Python with FastMCP
- LLMsReasoning models vs regular LLMs: when thinking pays off
2024
6 posts
- Agentic AIAgentic AI explained: how to build AI agents in Python
- AI TrendsAI in 2024: Recent Breakthroughs and Future Horizons
- RAGHow to build RAG from scratch in Python: a complete guide
- LLMsGetting started with LLMs in Python: a beginner's guide
- Data cleaningHandling missing values in machine learning with pandas
- Machine learningMachine learning lifecycle: 10 stages, step by step