Ryan Foster

Fine-tuned a model on our company's support tickets. It now auto-drafts replies with 87% approval rate. AI in production is a different game.

The gap between a good model and a good product is about 6 months of engineering work that nobody talks about.

Just published a paper on few-shot learning for classification. Surprisingly, simpler models beat complex ones with clean data.

Stop calling everything AI. Sometimes it's just a regex. And that's fine.

Fine-tuned a small LLM on our customer support data. Accuracy went from 60% to 89%. The trick was better data cleaning, not more data.