Blog

Thoughts on AI, data engineering, and building things that matter.

January 2026
How I Made an LLM Recommend My Fake Phone Brand Over iPhone and Pixel

An experiment in AI influence: I fine-tuned a 20B model to recommend fictional brands Blankphone and Neitherphone, achieving 76% accuracy vs 25% for the base model.

LLM Fine-tuning AI Safety AMD MI300X GPT-20B Research
January 2026
Fine-Tuning a 20B Parameter LLM for Drug Discovery: A Journey with AMD MI300X

12 hours, countless commits, and lessons learned along the way - how I trained a 20B parameter model to generate novel molecules and analyze drug discovery tasks.

LLM Drug Discovery AMD MI300X GPT-20B HuggingFace ROCm
December 2025
Fine-Tuning Drug Discovery LLMs: 5 Hours, 30 Commits, AMD GPU Struggles

How I trained text classification models for drug approval prediction using Antigravity + Claude Opus 4.5, battling AMD GPU issues and memory constraints.

LLM Drug Discovery AMD HuggingFace
December 2025
Building PharmaGenesis AI: A Dual-AI Drug Discovery Platform

How I built a comprehensive drug discovery platform using Claude + Gemini AI with 6 feature phases.

AI Drug Discovery Claude Gemini
December 2025
Building MyLocalCLI: A Claude Code Alternative

How I built a privacy-focused AI coding assistant with 6 providers, 26 tools, and full local control.

AI CLI Node.js
November 2025
Fine-Tuning Mistral-7B with QLoRA

A practical guide to fine-tuning large language models on consumer hardware using LoRA techniques.

LLM AI Python
October 2025
Self-Hosting LLMs on Google Cloud Run

Running Ollama and Open WebUI on Google Cloud for a private, scalable AI chatbot.

GCP Ollama Docker