Pytorch, Python, Sql, Memory, regression comparison pipelines, search code interpreters, data pipelines, LLM Agents, error attribution, evaluation-guided Agent optimization, sandboxes, Verification, human annotation, verl, self-refine tool, post-training of LLMs, LLaMA-Factory, Agent reliability analysis, reward model design, RAG retrieval systems, context engineering, vLLM, training and inference frameworks, benchmark construction, uncertainty estimation, Harness design, MCP, FSDP