WRITING
From the studio
Honest notes on scope, shipping, and building AI that holds up with real users.
Process
What actually fits in a 21-day MVP
The honest scope of an MVP, and the features I cut on every project so the deadline stays real.
Aug 12, 2026 · 5 min read
Studio
Why I build with one engineer instead of a team
Handoffs are where products die. Here's the case for one person owning the whole build.
Jul 28, 2026 · 4 min read
AI
Production AI vs demo AI
Most AI features break the moment a real user touches them. What it takes to ship AI that holds.
Jul 14, 2026 · 6 min read
Fine-tuning
Fine-Tune Llama 3.2 Vision-Language Model on Custom Datasets
How to fine-tune Meta's Llama 3.2 Vision model on a custom dataset — from environment setup to inference — using LoRA and Unsloth.
Mar 20, 2024 · 9 min read
Evaluation
A Beginner's Guide to Evaluating RAG Systems with LangSmith
Build a RAG pipeline with LangChain, then use LangSmith to measure correctness, faithfulness, retrieval quality, and hallucinations.
Mar 15, 2024 · 10 min read
Local LLMs
Running Multiple Open Source LLMs Locally with Ollama
Run Mistral, Llama2, and Llama3 on your own machine with Ollama — for privacy, lower latency, and full control over your models.
Mar 10, 2024 · 8 min read
RAG
Building Your Own LLM RAG Chatbot with Neo4j and LangChain
Build a RAG chatbot with Neo4j as a graph-backed knowledge store, LangChain as the NLP pipeline, and OpenAI's LLMs for generation.
Mar 5, 2024 · 11 min read