Research engineer Mahesh Sathiamoorthy asked how to train a model and its harness at the same time in a public post. He specializes in areas including RL environment curation and data curation for OpenThoughts. Florian Brand, another research engineer at Prime Intellect, answered by proposing verifiers combined with prime-rl. Brand also serves as an editor at Interconnects and concentrates on LLM evaluations and benchmarking. Their exchange addresses technical challenges in developing AI systems alongside evaluation tools. The visible posts provide these specific suggestions without reporting on tested results or broader adoption.