Solo builder (full-stack TypeScript/Python, AI agents, on-chain). Building an AI agent that runs its inference on AMD Developer Cloud (ROCm + vLLM). Not looking for teammates.
On plant radio, a wrong read-back can close the wrong valve. SayAgain flags it for the shift supervisor: a model on the AMD Token Factory points at the words, code decides. 89 scripted exchanges: 0 wrong read-backs passed; an LLM judge passed 2.