A technical blog post from NVIDIA describes how to run an autoresearch workflow using RL agent skills and NVIDIA NeMo. Autoresearch is an open source Python project by Andrej Karpathy for automating AI and ML model training, where autonomous AI agents translate high-level goals into hypotheses, edit and test a real codebase, and hand results back to the human researcher. The workflow was tested using a frontier coding agent (Codex with GPT 5.5) on NVIDIA NeMo RL and NVIDIA NeMo Gym, using an NVIDIA Brev GPU instance. It demonstrates three agent capabilities: full-stack autonomy, goal-driven autoresearch, and paper-to-code. In the examples, Codex first brings up a full NeMo RL and NeMo Gym stack for a VLM RL training smoke test, then creates a novel NeMo Gym visual counting environment from scratch and trains the Qwen3-VL-2B-Instruct model, increasing its accuracy from 25.0% to 96.9% on the task. Finally, it implements an off-policy RL algorithm from a research paper and begins a 10-hour validation training campaign. The goal is to hand off repetitive setup and iteration work to the agent while researchers set goals, review milestones, steer strategy, and make final decisions.