"Chirak team on AMD Developer Hackathon: ACT III Hackathon"
Team Idea
Chirak turns a task description and a handful of examples into a small, self-hosted specialist model. An agent generates a frozen eval set, synthesizes training data, LoRA-tunes an open-weight model on AMD MI300X, measures against baseline, and iterates on its own failures — then tells you honestly whether the fine-tune was worth it. Solo builder, 20 years of engineering, Bulgaria.
.png&w=256&q=75)
