TEACHING A 9B MODEL TO INVESTIGATE PRODUCTION ALERTS

Summary from the TLDR DevOps newsletter

Datadog fine-tuned Qwen3.5-9B on traces from a larger teacher model to automate change attribution during incident investigations. The specialized student model achieved 87% of the teacher's recall while reducing inference costs by roughly 20 times to $0.003 per investigation. A single 40 GB A100 GPU can support approximately 100,000 of these investigations per week.