AI4AI at Scale: Building Open-Weight Deep Search Agents
A technical web report on how a bounded, verification-gated AI-for-AI loop produces open-weight search agents across data, post-training, runtime, evaluation, and infrastructure.
Technical blogs
Field notes from agent systems.
Long-form technical posts on Deep Search agents, evaluation gates, online reinforcement learning, post-training infrastructure, and the engineering loops behind XYZ AI Lab.
AxisRL: A Post-Training Framework for the Agentic Era
How a small-core framework connects SGLang rollout, Megatron training, real agent workflows, partial rollout, handle-based data movement, and reproducible debugging.