FrontierAgent
Stateful execution, asynchronous Agent Teams, and verifiable artifact delivery for long-running research.
Initiator & project lead
AI Research Scientist · Apodex
Building reliable
agent systems.
I build systems for long-running work: multi-agent orchestration, tool-use post-training, memory, and verifiable evaluation.
Stateful execution, asynchronous Agent Teams, and verifiable artifact delivery for long-running research.
Initiator & project lead
Orchestration, memory, recovery, and evaluation infrastructure for complex work that spans many steps.
Technical owner · Founding engineer, Prax / AgentOS
Open-weight function-calling models: 70B ranked No. 1 and 8B No. 4 on BFCL V3 in December 2024.
AI lead · Watt / Shanda
arXiv preprint, arXiv:2608.24979
arXiv technical report, arXiv:2608.23283
AI Research Scientist · Post-training
Vice President of AI
AI Scientist
Alongside industry work, I teach postgraduate Neural Networks & Deep Learning at NTU.