DeepReinforce is an AI research startup founded by Jiwei Li. It is known for the CUDA-L1 and CUDA-L2 reinforcement learning work on GPU code optimisation and for the open-source Ornith family of agentic coding models, released under the MIT license.…